RAPHIA: A deep learning pipeline for the registration of MRI and whole-mount histopathology images of the prostate

Image registration can map the ground truth extent of prostate cancer from histopathology images onto MRI, facilitating the development of machine learning methods for early prostate cancer detection. Here, we present RAdiology PatHology Image Alignment (RAPHIA), an end-to-end pipeline for efficient and accurate registration of MRI and histopathology images. RAPHIA automates several time-consuming manual steps in existing approaches including prostate segmentation, estimation of the rotation angle and horizontal flipping in histopathology images, and estimation of MRI-histopathology slice correspondences. By utilizing deep learning registration networks, RAPHIA substantially reduces computational time. Furthermore, RAPHIA obviates the need for a multimodal image similarity metric by transferring histopathology image representations to MRI image representations and vice versa. With the assistance of RAPHIA, novice users achieved expert-level performance, and their mean error in estimating histopathology rotation angle was reduced by 51% (12 degrees vs 8 degrees), their mean accuracy of estimating histopathology flipping was increased by 5% (95.3% vs 100%), and their mean error in estimating MRI-histopathology slice correspondences was reduced by 45% (1.12 slices vs 0.62 slices). When compared to a recent conventional registration approach and a deep learning registration approach, RAPHIA achieved better mapping of histopathology cancer labels, with an improved mean Dice coefficient of cancer regions outlined on MRI and the deformed histopathology (0.44 vs 0.48 vs 0.50), and a reduced mean per-case processing time (51 vs 11 vs 4.5 min). The improved performance by RAPHIA allows efficient processing of large datasets for the development of machine learning models for prostate cancer detection on MRI. Our code is publicly available at: https://github.com/pimed/RAPHIA.

Towards optimal heterogeneous prostate radiotherapy dose prescriptions based on patient-specific or population-based biological features

BACKGROUND

Escalation of prescribed dose in prostate cancer (PCa) radiotherapy enables improvement in tumor control at the expense of increased toxicity. Opportunities for reduction of treatment toxicity may emerge if more efficient dose escalation can be achieved by redistributing the prescribed dose distribution according to the known heterogeneous, spatially-varying characteristics of the disease.

PURPOSE

To examine the potential benefits, limitations and characteristics of heterogeneous boost dose redistribution in PCa radiotherapy based on patient-specific and population-based spatial maps of tumor biological features.

METHOD

High-resolution prostate histology images, from a cohort of 63 patients, annotated with tumor location and grade, provided patient-specific "maps" and a population-based "atlas" of cell density and tumor probability. Dose prescriptions were derived for each patient based on a heterogeneous redistribution of the boost dose to the intraprostatic lesions, with the prescription maximizing patient tumor control probability (TCP). The impact on TCP was assessed under scenarios where the distribution of population-based biological data was ignored, partially included, or fully included in prescription generation. Heterogeneous dose prescriptions were generated for three combinations of maps and atlas, and for conventional fractionation (CF), extreme hypo-fractionation (EH), moderate hypo-fractionation (MH), and whole Pelvic RT + SBRT Boost (WPRT + SBRT). The predicted efficacy of the heterogeneous prescriptions was compared with equivalent homogeneous dose prescriptions.

RESULTS

TCPs for heterogeneous dose prescriptions were generally higher than those for homogeneous dose prescriptions. TCP escalation by heterogeneous dose prescription was the largest for CF. When only using population-based atlas data, the generated heterogeneous dose prescriptions of 55 to 58 patients (out of 63) had a higher TCP than for the corresponding homogeneous dose prescriptions. The TCPs of the heterogeneous dose prescriptions generated with the population-based atlas and tumor probability maps did not differ significantly from those using patient-specific biological information. The generated heterogeneous dose prescriptions achieved significantly higher TCP than homogeneous dose prescriptions in the posterior section of the prostate.

CONCLUSION

Heterogeneous dose prescriptions generated via biologically-optimized dose redistribution can produce higher TCP than the homogeneous dose prescriptions for the majority of the patients in the studied cohort. For scenarios where patient-specific biological information was unavailable or partially available, the generated heterogeneous dose prescriptions can still achieve TCP improvement relative to homogeneous dose prescriptions.

Robustness of magnetic resonance imaging and positron emission tomography radiomic features in prostate cancer: Impact on recurrence prediction after radiation therapy

Background and purpose

Radiomic features from MRI and PET are an emerging tool with potential to improve prostate cancer outcomes. However, feature robustness due to image segmentation variations is currently unknown. Therefore, this study aimed to evaluate the robustness of radiomic features with segmentation variations and their impact on predicting biochemical recurrence (BCR).

Materials and methods

Multi-scanner, pre-radiation therapy imaging from 142 patients with localised prostate cancer was used. Imaging included T2-weighted (T2), apparent diffusion coefficient (ADC) MRI, and prostate-specific membrane antigen (PSMA)-PET. The prostate gland and intraprostatic tumours were manually and automatically segmented, and differences were quantified using Dice Coefficient (DC). Radiomic features including shape, first-order, and texture features were extracted for each segmentation from original and filtered images. Intraclass Correlation Coefficient (ICC) and Mean Absolute Percentage Difference (MAPD) were used to assess feature robustness. Random forest (RF) models were developed for each segmentation using robust features to predict BCR.

Results

Prostate gland segmentations were more consistent (mean DC = 0.78) than tumour segmentations (mean DC = 0.46). 112 (3.6 %) radiomic features demonstrated 'excellent' robustness (ICC > 0.9 and MAPD < 1 %), and 480 features (15.4 %) demonstrated 'good' robustness (ICC > 0.75 and MAPD < 5 %). PET imaging provided more features with excellent robustness than T2 and ADC. RF models showed strong predictive power for BCR with a mean area under the receiver-operator-characteristics curve (AUC) of 0.89 (range 0.85-0.93).

Conclusion

When using radiomic features for predictive modelling, segmentation variability should be considered. To develop BCR predictive models, radiomic features from the entire prostate gland are preferable over tumour segmentation-based features.

Focal Boost in Prostate Cancer Radiotherapy: A Review of Planning Studies and Clinical Trials

BACKGROUND

Focal boost radiotherapy was developed to deliver elevated doses to functional sub-volumes within a target. Such a technique was hypothesized to improve treatment outcomes without increasing toxicity in prostate cancer treatment.

PURPOSE

To summarize and evaluate the efficacy and variability of focal boost radiotherapy by reviewing focal boost planning studies and clinical trials that have been published in the last ten years.

METHODS

Published reports of focal boost radiotherapy, that specifically incorporate dose escalation to intra-prostatic lesions (IPLs), were reviewed and summarized. Correlations between acute/late ≥G2 genitourinary (GU) or gastrointestinal (GI) toxicity and clinical factors were determined by a meta-analysis.

RESULTS

By reviewing and summarizing 34 planning studies and 35 trials, a significant dose escalation to the GTV and thus higher tumor control of focal boost radiotherapy were reported consistently by all reviewed studies. Reviewed trials reported a not significant difference in toxicity between focal boost and conventional radiotherapy. Acute ≥G2 GU and late ≥G2 GI toxicities were reported the most and least prevalent, respectively, and a negative correlation was found between the rate of toxicity and proportion of low-risk or intermediate-risk patients in the cohort.

CONCLUSION

Focal boost prostate cancer radiotherapy has the potential to be a new standard of care.

Parametric maps of spatial two-tissue compartment model for prostate dynamic contrast enhanced MRI - comparison with the standard tofts model in the diagnosis of prostate cancer

The spatial two-tissue compartment model (2TCM) was used to analyze prostate dynamic contrast enhanced (DCE) MRI data and compared with the standard Tofts model. A total of 29 patients with biopsy-confirmed prostate cancer were included in this IRB-approved study. MRI data were acquired on a Philips Achieva 3T-TX scanner. After T2-weighted and diffusion-weighted imaging, DCE data using 3D T1-FFE mDIXON sequence were acquired pre- and post-contrast media injection (0.1 mmol/kg Multihance) for 60 dynamic scans with temporal resolution of 8.3 s/image. The 2TCM has one fast ([Formula: see text] and [Formula: see text]) and one slow ([Formula: see text] and [Formula: see text]) exchanging compartment, compared with the standard Tofts model parameters (Ktrans and kep). On average, prostate cancer had significantly higher values (p < 0.01) than normal prostate tissue for all calculated parameters. There was a strong correlation (r = 0.94, p < 0.001) between Ktrans and [Formula: see text] for cancer, but weak correlation (r = 0.28, p < 0.05) between kep and [Formula: see text]. Average root-mean-square error (RMSE) in fits from the 2TCM was significantly smaller (p < 0.001) than the RMSE in fits from the Tofts model. Receiver operating characteristic (ROC) analysis showed that fast [Formula: see text] had the highest area under the curve (AUC) than any other individual parameter. The combined four parameters from the 2TCM had a considerably higher AUC value than the combined two parameters from the Tofts model. The 2TCM is useful for quantitative analysis of prostate DCE-MRI data and provides new information in the diagnosis of prostate cancer.

[68Ga]Ga-PSMA and [68Ga]Ga-RM2 PET/MRI vs. Histopathological Images in Prostate Cancer: A New Workflow for Spatial Co-Registration

This study proposed a new workflow for co-registering prostate PET images from a dual-tracer PET/MRI study with histopathological images of resected prostate specimens. The method aims to establish an accurate correspondence between PET/MRI findings and histology, facilitating a deeper understanding of PET tracer distribution and enabling advanced analyses like radiomics. To achieve this, images derived by three patients who underwent both [68Ga]Ga-PSMA and [68Ga]Ga-RM2 PET/MRI before radical prostatectomy were selected. After surgery, in the resected fresh specimens, fiducial markers visible on both histology and MR images were inserted. An ex vivo MRI of the prostate served as an intermediate step for co-registration between histological specimens and in vivo MRI examinations. The co-registration workflow involved five steps, ensuring alignment between histopathological images and PET/MRI data. The target registration error (TRE) was calculated to assess the precision of the co-registration. Furthermore, the DICE score was computed between the dominant intraprostatic tumor lesions delineated by the pathologist and the nuclear medicine physician. The TRE for the co-registration of histopathology and in vivo images was 1.59 mm, while the DICE score related to the site of increased intraprostatic uptake on [68Ga]Ga-PSMA and [68Ga]Ga-RM2 PET images was 0.54 and 0.75, respectively. This work shows an accurate co-registration method for histopathological and in vivo PET/MRI prostate examinations that allows the quantitative assessment of dual-tracer PET/MRI diagnostic accuracy at a millimetric scale. This approach may unveil radiotracer uptake mechanisms and identify new PET/MRI biomarkers, thus establishing the basis for precision medicine and future analyses, such as radiomics.

Detecting localised prostate cancer using radiomic features in PSMA PET and multiparametric MRI for biologically targeted radiation therapy

BACKGROUND

Prostate-Specific Membrane Antigen (PSMA) PET/CT and multiparametric MRI (mpMRI) are well-established modalities for identifying intra-prostatic lesions (IPLs) in localised prostate cancer. This study aimed to investigate the use of PSMA PET/CT and mpMRI for biologically targeted radiation therapy treatment planning by: (1) analysing the relationship between imaging parameters at a voxel-wise level and (2) assessing the performance of radiomic-based machine learning models to predict tumour location and grade.

METHODS

PSMA PET/CT and mpMRI data from 19 prostate cancer patients were co-registered with whole-mount histopathology using an established registration framework. Apparent Diffusion Coefficient (ADC) maps were computed from DWI and semi-quantitative and quantitative parameters from DCE MRI. Voxel-wise correlation analysis was conducted between mpMRI parameters and PET Standardised Uptake Value (SUV) for all tumour voxels. Classification models were built using radiomic and clinical features to predict IPLs at a voxel level and then classified further into high-grade or low-grade voxels.

RESULTS

Perfusion parameters from DCE MRI were more highly correlated with PET SUV than ADC or T2w. IPLs were best detected with a Random Forest Classifier using radiomic features from PET and mpMRI rather than either modality alone (sensitivity, specificity and area under the curve of 0.842, 0.804 and 0.890, respectively). The tumour grading model had an overall accuracy ranging from 0.671 to 0.992.

CONCLUSIONS

Machine learning classifiers using radiomic features from PSMA PET and mpMRI show promise for predicting IPLs and differentiating between high-grade and low-grade disease, which could be used to inform biologically targeted radiation therapy planning.

U-Net Architecture for Prostate Segmentation: The Impact of Loss Function on System Performance

Segmentation of the prostate gland from magnetic resonance images is rapidly becoming a standard of care in prostate cancer radiotherapy treatment planning. Automating this process has the potential to improve accuracy and efficiency. However, the performance and accuracy of deep learning models varies depending on the design and optimal tuning of the hyper-parameters. In this study, we examine the effect of loss functions on the performance of deep-learning-based prostate segmentation models. A U-Net model for prostate segmentation using T2-weighted images from a local dataset was trained and performance compared when using nine different loss functions, including: Binary Cross-Entropy (BCE), Intersection over Union (IoU), Dice, BCE and Dice (BCE + Dice), weighted BCE and Dice (W (BCE + Dice)), Focal, Tversky, Focal Tversky, and Surface loss functions. Model outputs were compared using several metrics on a five-fold cross-validation set. Ranking of model performance was found to be dependent on the metric used to measure performance, but in general, W (BCE + Dice) and Focal Tversky performed well for all metrics (whole gland Dice similarity coefficient (DSC): 0.71 and 0.74; 95HD: 6.66 and 7.42; Ravid 0.05 and 0.18, respectively) and Surface loss generally ranked lowest (DSC: 0.40; 95HD: 13.64; Ravid −0.09). When comparing the performance of the models for the mid-gland, apex, and base parts of the prostate gland, the models’ performance was lower for the apex and base compared to the mid-gland. In conclusion, we have demonstrated that the performance of a deep learning model for prostate segmentation can be affected by choice of loss function. For prostate segmentation, it would appear that compound loss functions generally outperform singles loss functions such as Surface loss.

Semi-quantitative and quantitative dynamic contrast-enhanced (DCE) MRI parameters as prostate cancer imaging biomarkers for biologically targeted radiation therapy

BACKGROUND

Biologically targeted radiation therapy treatment planning requires voxel-wise characterisation of tumours. Dynamic contrast enhanced (DCE) DCE MRI has shown promise in defining voxel-level biological characteristics. In this study we consider the relative value of qualitative, semi-quantitative and quantitative assessment of DCE MRI compared with diffusion weighted imaging (DWI) and T2-weighted (T2w) imaging to detect prostate cancer at the voxel level.

METHODS

Seventy prostate cancer patients had multiparametric MRI prior to radical prostatectomy, including T2w, DWI and DCE MRI. Apparent Diffusion Coefficient (ADC) maps were computed from DWI, and semi-quantitative and quantitative parameters computed from DCE MRI. Tumour location and grade were validated with co-registered whole mount histology. Kolmogorov-Smirnov tests were applied to determine whether MRI parameters in tumour and benign voxels were significantly different. Cohen's d was computed to quantify the most promising biomarkers. The Parker and Weinmann Arterial Input Functions (AIF) were compared for their ability to best discriminate between tumour and benign tissue. Classifier models were used to determine whether DCE MRI parameters improved tumour detection versus ADC and T2w alone.

RESULTS

All MRI parameters had significantly different data distributions in tumour and benign voxels. For low grade tumours, semi-quantitative DCE MRI parameter time-to-peak (TTP) was the most discriminating and outperformed ADC. For high grade tumours, ADC was the most discriminating followed by DCE MRI parameters Ktrans, the initial rate of enhancement (IRE), then TTP. Quantitative parameters utilising the Parker AIF better distinguished tumour and benign voxel values than the Weinmann AIF. Classifier models including DCE parameters versus T2w and ADC alone, gave detection accuracies of 78% versus 58% for low grade tumours and 85% versus 72% for high grade tumours.

CONCLUSIONS

Incorporating DCE MRI parameters with DWI and T2w gives improved accuracy for tumour detection at a voxel level. DCE MRI parameters should be used to spatially characterise tumour biology for biologically targeted radiation therapy treatment planning.

Challenges and Practical Solutions to MRI and Histology Matching and Measurements Using Available ImageJ Software Tools

Traditionally histology is the gold standard for the validation of imaging experiments. Matching imaging slices and histological sections and the precise outlining of corresponding tissue structures are difficult. Challenges are based on differences in imaging and histological slice thickness as well as tissue shrinkage and alterations after processing. Here we describe step-by-step instructions that might be used as a universal pathway to overlay MRI and histological images and for a correlation of measurements between imaging modalities. The free available (Fiji is just) ImageJ software tools were used for regions of interest transformation (ROIT) and alignment using a rat brain MRI as an example. The developed ROIT procedure was compared to a manual delineation of rat brain structures. The ROIT plugin was developed for ImageJ to enable an automatization of the image processing and structural analysis of the rodent brain.

A review of artificial intelligence in prostate cancer detection on imaging

A multitude of studies have explored the role of artificial intelligence (AI) in providing diagnostic support to radiologists, pathologists, and urologists in prostate cancer detection, risk-stratification, and management. This review provides a comprehensive overview of relevant literature regarding the use of AI models in (1) detecting prostate cancer on radiology images (magnetic resonance and ultrasound imaging), (2) detecting prostate cancer on histopathology images of prostate biopsy tissue, and (3) assisting in supporting tasks for prostate cancer detection (prostate gland segmentation, MRI-histopathology registration, MRI-ultrasound registration). We discuss both the potential of these AI models to assist in the clinical workflow of prostate cancer diagnosis, as well as the current limitations including variability in training data sets, algorithms, and evaluation criteria. We also discuss ongoing challenges and what is needed to bridge the gap between academic research on AI for prostate cancer and commercial solutions that improve routine clinical care.

A statistical, voxelised model of prostate cancer for biologically optimised radiotherapy

Background and purpose

Radiation therapy (RT) is commonly indicated for treatment of prostate cancer (PC). Biologicallyoptimised RT for PC may improve disease-free survival. This requires accurate spatial localisation and characterisation of tumour lesions. We aimed to generate a statistical, voxelised biological model to complement in vivomultiparametric MRI data to facilitate biologically-optimised RT.

Material and methods

Ex vivo prostate MRI and histopathological imaging were acquired for 63 PC patients. These data were co-registered to derive three-dimensional distributions of graded tumour lesions and cell density. Novel registration processes were used to map these data to a common reference geometry. Voxelised statistical models of tumour probability and cell density were generated to create the PC biological atlas. Cell density models were analysed using the Kullback–Leibler divergence to compare normal vs. lognormal approximations to empirical data.

Results

A reference geometry was constructed using ex vivo MRI space, patient data were deformably registered using a novel anatomy-guided process. Substructure correspondence was maintained using peripheral zone definitions to address spatial variability in prostate anatomy between patients. Three distinct approaches to interpolation were designed to map contours, tumour annotations and cell density maps from histology into ex vivo MRI space. Analysis suggests a log-normal model provides a more consistent representation of cell density when compared to a linear-normal model.

Conclusion

A biological model has been created that combines spatial distributions of tumour characteristics from a population into three-dimensional, voxelised, statistical models. This tool will be used to aid the development of biologically-optimised RT for PC patients.

Correlation of in-vivo imaging with histopathology: A review

Despite tremendous advancements in in vivo imaging modalities, there remains substantial uncertainty with respect to tumor delineation on in these images. Histopathology remains the gold standard for determining the extent of malignancy, with in vivo imaging to histopathologic correlation enabling spatial comparisons. In this review, the steps necessary for successful imaging to histopathologic correlation are described, including in vivo imaging, resection, fixation, specimen sectioning (sectioning technique, securing technique, orientation matching, slice matching), microtome sectioning and staining, correlation (including image registration) and performance evaluation. The techniques used for each of these steps are also discussed. Hundreds of publications from the past 20 years were surveyed, and 62 selected for detailed analysis. For these 62 publications, each stage of the correlative pathology process (and the sub-steps of specimen sectioning) are listed. A statistical analysis was conducted based on 19 studies that reported target registration error as their performance metric. While some methods promise greater accuracy, they may be expensive. Due to the complexity of the processes involved, correlative pathology studies generally include a small number of subjects, which hinders advanced developments in this field.

Biologically Targeted Radiation Therapy: Incorporating Patient-Specific Hypoxia Data Derived from Quantitative Magnetic Resonance Imaging

PURPOSE

Hypoxia has been linked to radioresistance. Strategies to safely dose escalate dominant intraprostatic lesions have shown promising results, but further dose escalation to overcome the effects of hypoxia require a novel approach to constrain the dose in normal tissue.to safe levels. In this study, we demonstrate a biologically targeted radiotherapy (BiRT) approach that can utilise multiparametric magnetic resonance imaging (mpMRI) to target hypoxia for favourable treatment outcomes.

METHODS

mpMRI-derived tumour biology maps, developed via a radiogenomics study, were used to generate individualised, hypoxia-targeting prostate IMRT plans using an ultra- hypofractionation schedule. The spatial distribution of mpMRI textural features associated with hypoxia-related genetic profiles was used as a surrogate of tumour hypoxia. The effectiveness of the proposed approach was assessed by quantifying the potential benefit of a general focal boost approach on tumour control probability, and also by comparing the dose to organs at risk (OARs) with hypoxia-guided focal dose escalation (DE) plans generated for five patients.

RESULTS

Applying an appropriately guided focal boost can greatly mitigate the impact of hypoxia. Statistically significant reductions in rectal and bladder dose were observed for hypoxia-targeting, biologically optimised plans compared to isoeffective focal DE plans.

CONCLUSION

Results of this study suggest the use of mpMRI for voxel-level targeting of hypoxia, along with biological optimisation, can provide a mechanism for guiding focal DE that is considerably more efficient than application of a general, dose-based optimisation, focal boost.

Technical Note: Histological validation of anatomical imaging for breast modeling using a novel cryo-microtome

PURPOSE

Precise correlation between three-dimensional (3D) imaging and histology can aid biomechanical modeling of the breast. We develop a framework to register ex vivo images to histology using a novel cryo-fluorescence tomography (CFT) device.

METHODS

A formalin-fixed cadaveric breast specimen, including chest wall, was subjected to high-resolution magnetic resonance (MR) imaging. The specimen was then frozen and embedded in an optimal cutting temperature (OCT) compound. The OCT block was placed in a CFT device with an overhead camera and 50 μm thick slices were successively shaved off the block. After each shaving, the block-face was photographed. At select locations including connective/adipose tissue, muscle, skin, and fibroglandular tissue, 20 μm sections were transferred onto cryogenic tape for manual hematoxylin and eosin staining, histological assessment, and image capture. A 3D white-light image was automatically reconstructed from the photographs by aligning fiducial markers embedded in the OCT block. The 3D MR image, 3D white-light image, and photomicrographs were rigidly registered. Target registration errors (TREs) were computed based on 10 pairs of points marked at fibroglandular intersections. The overall MR-histology registration was used to compare the MR intensities at tissue extraction sites with a one-way analysis of variance.

RESULTS

The MR image to CFT-captured white-light image registration achieved a mean TRE of 0.73 ± 0.25 mm (less than the 1 mm MR slice resolution). The block-face white-light image and block-face photomicrograph registration showed visually indistinguishable alignment of anatomical structures and tissue boundaries. The MR intensities at the four tissue sites identified from histology differed significantly (p < 0.01). Each tissue pair, except the skin-connective/adipose tissue pair, also had significantly different MR intensities (p < 0.01).

CONCLUSIONS

Fine sectioning in a highly controlled imaging/sectioning environment enables accurate registration between the MR image and histology. Statistically significant differences in MR signal intensities between histological tissues are indicators for the specificity of correlation between MRI and histology.

Registration of histopathology to magnetic resonance imaging of prostate cancer

BACKGROUND AND PURPOSE

The diagnostic accuracy of new imaging techniques requires validation, preferably by histopathological verification. The aim of this study was to develop and present a registration procedure between histopathology and in-vivo magnetic resonance imaging (MRI) of the prostate, to estimate its uncertainty and to evaluate the benefit of adding a contour-correcting registration.

MATERIALS AND METHODS

For twenty-five prostate cancer patients, planned for radical prostatectomy, a 3D-printed prostate mold based on in-vivo MRI was created and an ex-vivo MRI of the specimen, placed inside the mold, was performed. Each histopathology slice was registered to its corresponding ex-vivo MRI slice using a 2D-affine registration. The ex-vivo MRI was rigidly registered to the in-vivo MRI and the resulting transform was applied to the histopathology stack. A 2D deformable registration was used to correct for specimen distortion concerning the specimen's fit inside the mold. We estimated the spatial uncertainty by comparing positions of landmarks in the in-vivo MRI and the corresponding registered histopathology stack.

RESULTS

Eighty-four landmarks were identified, located in the urethra (62%), prostatic cysts (33%), and the ejaculatory ducts (5%). The median number of landmarks was 3 per patient. We showed a median in-plane error of 1.8 mm before and 1.7 mm after the contour-correcting deformable registration. In patients with extraprostatic margins, the median in-plane error improved from 2.1 mm to 1.8 mm after the contour-correcting deformable registration.

CONCLUSIONS

Our registration procedure accurately registers histopathology to in-vivo MRI, with low uncertainty. The contour-correcting registration was beneficial in patients with extraprostatic surgical margins.

Standard versus hypofractionated intensity-modulated radiotherapy for prostate cancer: assessing the impact on dose modulation and normal tissue effects when using patient-specific cancer biology

Hypofractionation of prostate cancer radiotherapy achieves tumour control at lower total radiation doses, however, increased rectal and bladder toxicities have been observed. To realise the radiobiological advantage of hypofractionation whilst minimising harm, the potential reduction in dose to organs at risk was investigated for biofocused radiotherapy. Patient-specific tumour location and cell density information were derived from multiparametric imaging. Uniform-dose plans and biologically-optimised plans were generated for a standard schedule (78 Gy/39 fractions) and hypofractionated schedules (60 Gy/20 fractions and 36.25 Gy/5 fractions). Results showed that biologically-optimised plans yielded statistically lower doses to the rectum and bladder compared to isoeffective uniform-dose plans for all fractionation schedules. A reduction in the number of fractions increased the target dose modulation required to achieve equal tumour control. On average, biologically-optimised, moderately-hypofractionated plans demonstrated 15.3% (p-value: <0.01) and 23.8% (p-value: 0.02) reduction in rectal and bladder dose compared with standard fractionation. The tissue-sparing effect was more pronounced in extreme hypofractionation with mean reduction in rectal and bladder dose of 43.3% (p-value: < 0.01) and 41.8% (p-value: 0.02), respectively. This study suggests that the ability to utilise patient-specific tumour biology information will provide greater incentive to employ hypofractionation in the treatment of localised prostate cancer with radiotherapy. However, to exploit the radiobiological advantages given by hypofractionation, greater attention to geometric accuracy is required due to increased sensitivity to treatment uncertainties.

ProsRegNet: A deep learning framework for registration of MRI and histopathology images of the prostate

Magnetic resonance imaging (MRI) is an increasingly important tool for the diagnosis and treatment of prostate cancer. However, interpretation of MRI suffers from high inter-observer variability across radiologists, thereby contributing to missed clinically significant cancers, overdiagnosed low-risk cancers, and frequent false positives. Interpretation of MRI could be greatly improved by providing radiologists with an answer key that clearly shows cancer locations on MRI. Registration of histopathology images from patients who had radical prostatectomy to pre-operative MRI allows such mapping of ground truth cancer labels onto MRI. However, traditional MRI-histopathology registration approaches are computationally expensive and require careful choices of the cost function and registration hyperparameters. This paper presents ProsRegNet, a deep learning-based pipeline to accelerate and simplify MRI-histopathology image registration in prostate cancer. Our pipeline consists of image preprocessing, estimation of affine and deformable transformations by deep neural networks, and mapping cancer labels from histopathology images onto MRI using estimated transformations. We trained our neural network using MR and histopathology images of 99 patients from our internal cohort (Cohort 1) and evaluated its performance using 53 patients from three different cohorts (an additional 12 from Cohort 1 and 41 from two public cohorts). Results show that our deep learning pipeline has achieved more accurate registration results and is at least 20 times faster than a state-of-the-art registration algorithm. This important advance will provide radiologists with highly accurate prostate MRI answer keys, thereby facilitating improvements in the detection of prostate cancer on MRI. Our code is freely available at https://github.com/pimed//ProsRegNet.

3D Registration of pre-surgical prostate MRI and histopathology images via super-resolution volume reconstruction

The use of MRI for prostate cancer diagnosis and treatment is increasing rapidly. However, identifying the presence and extent of cancer on MRI remains challenging, leading to high variability in detection even among expert radiologists. Improvement in cancer detection on MRI is essential to reducing this variability and maximizing the clinical utility of MRI. To date, such improvement has been limited by the lack of accurately labeled MRI datasets. Data from patients who underwent radical prostatectomy enables the spatial alignment of digitized histopathology images of the resected prostate with corresponding pre-surgical MRI. This alignment facilitates the delineation of detailed cancer labels on MRI via the projection of cancer from histopathology images onto MRI. We introduce a framework that performs 3D registration of whole-mount histopathology images to pre-surgical MRI in three steps. First, we developed a novel multi-image super-resolution generative adversarial network (miSRGAN), which learns information useful for 3D registration by producing a reconstructed 3D MRI. Second, we trained the network to learn information between histopathology slices to facilitate the application of 3D registration methods. Third, we registered the reconstructed 3D histopathology volumes to the reconstructed 3D MRI, mapping the extent of cancer from histopathology images onto MRI without the need for slice-to-slice correspondence. When compared to interpolation methods, our super-resolution reconstruction resulted in the highest PSNR relative to clinical 3D MRI (32.15 dB vs 30.16 dB for BSpline interpolation). Moreover, the registration of 3D volumes reconstructed via super-resolution for both MRI and histopathology images showed the best alignment of cancer regions when compared to (1) the state-of-the-art RAPSODI approach, (2) volumes that were not reconstructed, or (3) volumes that were reconstructed using nearest neighbor, linear, or BSpline interpolations. The improved 3D alignment of histopathology images and MRI facilitates the projection of accurate cancer labels on MRI, allowing for the development of improved MRI interpretation schemes and machine learning models to automatically detect cancer on MRI.

Histological Validation of MRI: A Review of Challenges in Registration of Imaging and Whole-Mount Histopathology

Rigorous validation with ground truth information such as histology is needed to reliably assess the current and potential value of MRI techniques to characterize tissue and identify disease-related tissue alterations. Commonly used methods that aim to directly correlate histology and MRI data generally fall short of this goal due to spatial errors that preclude direct matching. Errors result from tissue deformation, differences in spatial resolution and slice thickness, non-coplanar and/or nonintersecting plane orientations, and different image contrast mechanisms. Some of these problems arise from limitations in standard protocols for clinical tissue processing and histology-based pathology reporting, and to some extent can be addressed by modifications to standard protocols without compromising the clinical process. Typical modifications include ex vivo specimen MRI, block-face photography, addition of fiducial markers, and 3D printed molds to constrain tissue deformation and guide sectioning. This review summarizes the advantages and limitations of MRI validation techniques based on coregistration of MRI with whole-mount histology of tissue specimens. LEVEL OF EVIDENCE: 4 TECHNICAL EFFICACY STAGE: 1.

Registration of presurgical MRI and histopathology images from radical prostatectomy via RAPSODI

PURPOSE

Magnetic resonance imaging (MRI) has great potential to improve prostate cancer diagnosis; however, subtle differences between cancer and confounding conditions render prostate MRI interpretation challenging. The tissue collected from patients who undergo radical prostatectomy provides a unique opportunity to correlate histopathology images of the prostate with preoperative MRI to accurately map the extent of cancer from histopathology images onto MRI. We seek to develop an open-source, easy-to-use platform to align presurgical MRI and histopathology images of resected prostates in patients who underwent radical prostatectomy to create accurate cancer labels on MRI.

METHODS

Here, we introduce RAdiology Pathology Spatial Open-Source multi-Dimensional Integration (RAPSODI), the first open-source framework for the registration of radiology and pathology images. RAPSODI relies on three steps. First, it creates a three-dimensional (3D) reconstruction of the histopathology specimen as a digital representation of the tissue before gross sectioning. Second, RAPSODI registers corresponding histopathology and MRI slices. Third, the optimized transforms are applied to the cancer regions outlined on the histopathology images to project those labels onto the preoperative MRI.

RESULTS

We tested RAPSODI in a phantom study where we simulated various conditions, for example, tissue shrinkage during fixation. Our experiments showed that RAPSODI can reliably correct multiple artifacts. We also evaluated RAPSODI in 157 patients from three institutions that underwent radical prostatectomy and have very different pathology processing and scanning. RAPSODI was evaluated in 907 corresponding histpathology-MRI slices and achieved a Dice coefficient of 0.97 ± 0.01 for the prostate, a Hausdorff distance of 1.99 ± 0.70 mm for the prostate boundary, a urethra deviation of 3.09 ± 1.45 mm, and a landmark deviation of 2.80 ± 0.59 mm between registered histopathology images and MRI.

CONCLUSION

Our robust framework successfully mapped the extent of cancer from histopathology slices onto MRI providing labels from training machine learning methods to detect cancer on MRI.

Voxel-level biological optimisation of prostate IMRT using patient-specific tumour location and clonogen density derived from mpMRI

AIMS

This study aimed to develop a framework for optimising prostate intensity-modulated radiotherapy (IMRT) based on patient-specific tumour biology, derived from multiparametric MRI (mpMRI). The framework included a probabilistic treatment planning technique in the effort to yield dose distributions with an improved expected treatment outcome compared with uniform-dose planning approaches.

METHODS

IMRT plans were generated for five prostate cancer patients using two inverse planning methods: uniform-dose to the planning target volume and probabilistic biological optimisation for clinical target volume tumour control probability (TCP) maximisation. Patient-specific tumour location and clonogen density information were derived from mpMRI and geometric uncertainties were incorporated in the TCP calculation. Potential reduction in dose to sensitive structures was assessed by comparing dose metrics of uniform-dose plans with biologically-optimised plans of an equivalent level of expected tumour control.

RESULTS

The planning study demonstrated biological optimisation has the potential to reduce expected normal tissue toxicity without sacrificing local control by shaping the dose distribution to the spatial distribution of tumour characteristics. On average, biologically-optimised plans achieved 38.6% (p-value: < 0.01) and 51.2% (p-value: < 0.01) reduction in expected rectum and bladder equivalent uniform dose, respectively, when compared with uniform-dose planning.

CONCLUSIONS

It was concluded that varying the dose distribution within the prostate to take account for each patient's clonogen distribution was feasible. Lower doses to normal structures compared to uniform-dose plans was possible whilst providing robust plans against geometric uncertainties. Further validation in a larger cohort is warranted along with considerations for adaptive therapy and limiting urethral dose.

Association analysis between quantitative MRI features and hypoxia-related genetic profiles in prostate cancer: a pilot study

OBJECTIVE

To investigate the association between multiparametric MRI (mpMRI) imaging features and hypoxia-related genetic profiles in prostate cancer.

METHODS

In vivo mpMRI was acquired from six patients prior to radical prostatectomy. Sequences included T₂ weighted (T₂W) imaging, diffusion-weighted imaging, dynamic contrast enhanced MRI and blood oxygen-level dependent imaging. Imaging data were co-registered with histology using three-dimensional deformable registration methods. Texture features were extracted from T₂W images and parametric maps from functional MRI. Full transcriptome genetic profiles were obtained using next generation sequencing from the prostate specimens. Pearson correlation coefficients were calculated between mpMRI data and hypoxia-related gene expression levels. Results were validated using glucose transporter one immunohistochemistry (IHC).

RESULTS

Correlation analysis identified 34 candidate imaging features (six from the mpMRI data and 28 from T₂W texture features). The IHC validation showed that 16 out of the 28 T₂W texture features achieved weak but significant correlations (p < 0.05).

CONCLUSIONS

Weak associations between mpMRI features and hypoxia gene expressions were found. This indicates the potential use of MRI in assessing hypoxia status in prostate cancer. Further validation is required due to the low correlation levels.

ADVANCES IN KNOWLEDGE

This is a pilot study using radiogenomics approaches to address hypoxia within the prostate, which provides an opportunity for hypoxia-guided selective treatment techniques.

Correlation of ultra-high field MRI with histopathology for evaluation of rectal cancer heterogeneity

Current clinical MRI techniques in rectal cancer have limited ability to examine cancer stroma. The differentiation of tumour from desmoplasia or fibrous tissue remains a challenge. Standard MRI cannot differentiate stage T1 from T2 (invasion of muscularis propria) tumours. Diffusion tensor imaging (DTI) can probe tissue structure and organisation (anisotropy). The purpose of this study was to examine DTI-MRI derived imaging markers of rectal cancer stromal heterogeneity and tumour extent ex vivo. DTI-MRI at ultra-high magnetic field (11.7 tesla) was used to examine the stromal microstructure of malignant and normal rectal tissue ex vivo, and the findings were correlated with histopathology. Images obtained from DTI-MRI (A0, apparent diffusion coefficient and fractional anisotropy (FA)) were used to probe rectal cancer stromal heterogeneity. FA provided the best discrimination between cancer and desmoplasia, fibrous tissue and muscularis propria. Cancer had relatively isotropic diffusion (mean FA 0.14), whereas desmoplasia (FA 0.31) and fibrous tissue (FA 0.34) had anisotropic diffusion with significantly higher FA than cancer (p < 0.001). Tumour was distinguished from muscularis propria (FA 0.61) which was highly anisotropic with higher FA than cancer (p < 0.001). This study showed that DTI-MRI can assist in more accurately defining tumour extent in rectal cancer.

Multiparametric MRI and radiomics in prostate cancer: a review

Multiparametric MRI (mpMRI) is an imaging modality that combines anatomical MR imaging with one or more functional MRI sequences. It has become a versatile tool for detecting and characterising prostate cancer (PCa). The traditional role of mpMRI was confined to PCa staging, but due to the advanced imaging techniques, its role has expanded to various stages in clinical practises including tumour detection, disease monitor during active surveillance and sequential imaging for patient follow-up. Meanwhile, with the growing speed of data generation and the increasing volume of imaging data, it is highly demanded to apply computerised methods to process mpMRI data and extract useful information. Hence quantitative analysis for imaging data using radiomics has become an emerging paradigm. The application of radiomics approaches in prostate cancer has not only enabled automatic localisation of the disease but also provided a non-invasive solution to assess tumour biology (e.g. aggressiveness and the presence of hypoxia). This article reviews mpMRI and its expanding role in PCa detection, staging and patient management. Following that, an overview of prostate radiomics will be provided, with a special focus on its current applications as well as its future directions.

Voxel-wise prostate cell density prediction using multiparametric magnetic resonance imaging and machine learning

BACKGROUND

There are currently no methods to estimate cell density in the prostate. This study aimed to develop predictive models to estimate prostate cell density from multiparametric magnetic resonance imaging (mpMRI) data at a voxel level using machine learning techniques.

MATERIAL AND METHODS

In vivo mpMRI data were collected from 30 patients before radical prostatectomy. Sequences included T2-weighted imaging, diffusion-weighted imaging and dynamic contrast-enhanced imaging. Ground truth cell density maps were computed from histology and co-registered with mpMRI. Feature extraction and selection were performed on mpMRI data. Final models were fitted using three regression algorithms including multivariate adaptive regression spline (MARS), polynomial regression (PR) and generalised additive model (GAM). Model parameters were optimised using leave-one-out cross-validation on the training data and model performance was evaluated on test data using root mean square error (RMSE) measurements.

RESULTS

Predictive models to estimate voxel-wise prostate cell density were successfully trained and tested using the three algorithms. The best model (GAM) achieved a RMSE of 1.06 (± 0.06) × 10³ cells/mm² and a relative deviation of 13.3 ± 0.8%.

CONCLUSION

Prostate cell density can be quantitatively estimated non-invasively from mpMRI data using high-quality co-registered data at a voxel level. These cell density predictions could be used for tissue classification, treatment response evaluation and personalised radiotherapy.

A system using patient-specific 3D-printed molds to spatially align in vivo MRI with ex vivo MRI and whole-mount histopathology for prostate cancer research

BACKGROUND

Patient-specific 3D-printed molds and ex vivo MRI of the resected prostate have been two important strategies to align MRI with whole-mount histopathology (WMHP) for prostate cancer (PCa) research, but the combination of these two strategies has not been systematically evaluated.

PURPOSE

To develop and evaluate a system that combines patient-specific 3D-printed molds with ex vivo MRI (ExV) to spatially align in vivo MRI (InV), ExV, and WMHP in PCa patients.

STUDY TYPE

Prospective cohort study.

POPULATION

Seventeen PCa patients who underwent 3T MRI and robotic-assisted laparoscopic radical prostatectomy (RALP).

FIELD STRENGTH/SEQUENCES

T₂ -weighted turbo spin-echo sequences at 3T.

ASSESSMENT

Immediately after RALP, the fresh whole prostate specimens were imaged in patient-specific 3D-printed molds by 3T MRI and then sectioned to create WMHP slides. The time required for ExV was measured to assess impact on workflow. InV, ExV, and WMHP images were registered. Spatial alignment was evaluated using: slide offset (mm) between ExV slice locations and WMHP slides; overlap of the 3D prostate contour on InV versus ExV using Dice's coefficient (0 to 1); and 2D target registration error (TRE, mm) between corresponding landmarks on InV, ExV, and WMHP. Data are reported as mean ± standard deviation (SD).

STATISTICAL TESTING

Differences in 2D TRE before versus after registration were compared using the Wilcoxon signed-rank test (P < 0.05 considered significant).

RESULTS

ExV (duration 115 ± 15 min) was successfully incorporated into the workflow for all cases. Absolute slide offset was 1.58 ± 1.57 mm. Dice's coefficient was 0.865 ± 0.035. 2D TRE was significantly reduced after registration (P < 0.01) with mean (±SD of per patient means) of 1.9 ± 0.6 mm for InV versus ExV, 1.4 ± 0.5 mm for WMHP versus ExV, and 2.0 ± 0.5 mm for WMHP versus InV.

DATA CONCLUSION

The proposed system combines patient-specific 3D-printed molds with ExV to achieve spatial alignment between InV, ExV, and WMHP with mean 2D TRE of 1-2 mm.

LEVEL OF EVIDENCE

2 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019;49:270-279.

Accurate validation of ultrasound imaging of prostate cancer: a review of challenges in registration of imaging and histopathology

As the development of modalities for prostate cancer (PCa) imaging advances, the challenge of accurate registration between images and histopathologic ground truth becomes more pressing. Localization of PCa, rather than detection, requires a pixel-to-pixel validation of imaging based on histopathology after radical prostatectomy. Such a registration procedure is challenging for ultrasound modalities; not only the deformations of the prostate after resection have to be taken into account, but also the deformation due to the employed transrectal probe and the mismatch in orientation between imaging planes and pathology slices. In this work, we review the latest techniques to facilitate accurate validation of PCa localization in ultrasound imaging studies and extrapolate a general strategy for implementation of a registration procedure.

Registration Accuracy of Patient-Specific, Three-Dimensional-Printed Prostate Molds for Correlating Pathology With Magnetic Resonance Imaging

OBJECTIVE

This investigation was performed to evaluate the registration accuracy between magnetic resonance imaging (MRI) and pathology using three-dimensional (3-D) printed molds.

METHODS

Tissue-mimicking prostate phantoms were manufactured with embedded fiducials. The fiducials were used to measure and compare target registration error (TRE) between phantoms that were sliced by hand versus phantoms that were sliced within 3-D-printed molds. Subsequently, ten radical prostatectomy specimens were placed inside molds, scanned with MRI, and then sliced. The ex vivo scan was used to assess the true location of whole mount (WM) slides relative to in vivo MRI. The TRE between WM and in vivo MRI was measured using anatomic landmarks.

RESULTS

Manually sliced phantoms had a 4.1-mm mean TRE, whereas mold-sliced phantoms had a 1.9-mm mean TRE. Similarly, mold-assisted slicing reduced mean angular misalignment around the left-right (LR) anatomic axis from 10.7° to 4.5°. However, ex vivo MRI revealed that excised prostates were misaligned within molds, including a mean 14° rotation about the LR axis. The mean in-plane TRE was 3.3 mm using molds alone and 2.2 mm after registration was corrected with ex vivo MRI.

CONCLUSION

Patient-specific molds improved accuracy relative to manual slicing techniques in a phantom model. However, the registration accuracy of surgically resected specimens was limited by their imperfect fit within molds. This limitation can be overcome with the addition of ex vivo imaging.

SIGNIFICANCE

The accuracy of 3-D-printed molds was characterized, quantifying their utility for facilitating MRI-pathology registration.

Focal therapy for prostate cancer: the technical challenges

Focal therapy for prostate cancer has been proposed as an alternative treatment to whole gland therapy, offering the opportunity for tumor dose escalation and/or reduced toxicity. Brachytherapy, either low-dose-rate or high-dose-rate, provides an ideal approach, offering both precision in dose delivery and opportunity for a highly conformal, non-uniform dose distribution. Whilst multiple consensus documents have published clinical guidelines for patient selection, there are insufficient data to provide clear guidelines on target volume delineation, treatment planning margins, treatment planning approaches, and many other technical issues that should be considered before implementing a focal brachytherapy program. Without consensus guidelines, there is the potential for a diversity of practices to develop, leading to challenges in interpreting outcome data from multiple centers. This article provides an overview of the technical considerations for the implementation of a clinical service, and discusses related topics that should be considered in the design of clinical trials to ensure precise and accurate methods are applied for focal brachytherapy treatments.

Predicting prostate tumour location from multiparametric MRI using Gaussian kernel support vector machines: a preliminary study

The performance of a support vector machine (SVM) algorithm was investigated to predict prostate tumour location using multi-parametric MRI (mpMRI) data. The purpose was to obtain information of prostate tumour location for the implementation of bio-focused radiotherapy. In vivo mpMRI data were collected from 16 patients prior to radical prostatectomy. Sequences included T2-weighted imaging, diffusion-weighted imaging and dynamic contrast enhanced imaging. In vivo mpMRI was registered with 'ground truth' histology, using ex vivo MRI as an intermediate registration step to improve accuracy. Prostate contours were delineated by a radiation oncologist and tumours were annotated on histology by a pathologist. Five patients with minimal imaging artefacts were selected for this study. A Gaussian kernel SVM was trained and tested on different patient data subsets. Parameters were optimised using leave-oneout cross validation. Signal intensities of mpMRI were used as features and histology annotations as true labels. Prediction accuracy, as well as area under the curve (AUC) of the receiver operating characteristics (ROC) curve, were used to assess performance. Results demonstrated the prediction accuracy ranged from 70.4 to 87.1% and AUC of ROC ranged from 0.81 to 0.94. Additional investigations showed the apparent diffusion coefficient map from diffusion weighted imaging was the most important imaging modality for predicting tumour location. Future work will incorporate additional patient data into the framework to increase the sensitivity and specificity of the model, and will be extended to incorporate predictions of biological characteristics of the tumour which will be used in bio-focused radiotherapy optimisation.

PET and MRI: Is the Whole Greater than the Sum of Its Parts?

Over the past decades, imaging in oncology has been undergoing a "quiet" revolution to treat images as data, not as pictures. This revolution has been sparked by technological advances that enable capture of images that reflect not only anatomy, but also of tissue metabolism and physiology in situ Important advances along this path have been the increasing power of MRI, which can be used to measure spatially dependent differences in cell density, tissue organization, perfusion, and metabolism. In parallel, PET imaging allows quantitative assessment of the spatial localization of positron-emitting compounds, and it has also been constantly improving in the number of imageable tracers to measure metabolism and expression of macromolecules. Recent years have witnessed another technological advance, wherein these two powerful modalities have been physically merged into combined PET/MRI systems, appropriate for both preclinical or clinical imaging. As with all new enabling technologies driven by engineering physics, the full extent of potential applications is rarely known at the outset. In the work of Schmitz and colleagues, the authors have combined multiparametric MRI and PET imaging to address the important issue of intratumoral heterogeneity in breast cancer using both preclinical and clinical data. With combined PET and MRI and sophisticated machine-learning tools, they have been able identify multiple coexisting regions ("habitats") within living tumors and, in some cases, have been able to assign these habitats to known histologies. This work addresses an issue of fundamental importance to both cancer biology and cancer care. As with most new paradigm-shifting applications, it is not the last word on the subject and introduces a number of new avenues of investigation to pursue. Cancer Res; 76(21); 6163-6. ©2016 AACR.

Magnetic Resonance Imaging Underestimation of Prostate Cancer Geometry: Use of Patient Specific Molds to Correlate Images with Whole Mount Pathology

PURPOSE

We evaluated the accuracy of magnetic resonance imaging in determining the size and shape of localized prostate cancer.

MATERIALS AND METHODS

The subjects were 114 men who underwent multiparametric magnetic resonance imaging before radical prostatectomy with patient specific mold processing of the specimen from 2013 to 2015. T2-weighted images were used to contour the prostate capsule and cancer suspicious regions of interest. The contours were used to design and print 3-dimensional custom molds, which permitted alignment of excised prostates with magnetic resonance imaging scans. Tumors were reconstructed in 3 dimensions from digitized whole mount sections. Tumors were then matched with regions of interest and the relative geometries were compared.

RESULTS

Of the 222 tumors evident on whole mount sections 118 had been identified on magnetic resonance imaging. For the 118 regions of interest mean volume was 0.8 cc and the longest 3-dimensional diameter was 17 mm. However, for matched pathological tumors, of which most were Gleason score 3 + 4 or greater, mean volume was 2.5 cc and the longest 3-dimensional diameter was 28 mm. The median tumor had a 13.5 mm maximal extent beyond the magnetic resonance imaging contour and 80% of cancer volume from matched tumors was outside region of interest boundaries. Size estimation was most accurate in the axial plane and least accurate along the base-apex axis.

CONCLUSIONS

Magnetic resonance imaging consistently underestimates the size and extent of prostate tumors. Prostate cancer foci had an average diameter 11 mm longer and a volume 3 times greater than T2-weighted magnetic resonance imaging segmentations. These results may have important implications for the assessment and treatment of prostate cancer.