Influence of image contrasts and reconstruction methods on the classification of multiple sclerosis-like lesions in simulated sodium magnetic resonance imaging

PURPOSE

To evaluate the classifiability of small multiple sclerosis (MS)-like lesions in simulated sodium (²³ Na) MRI for different ²³ Na MRI contrasts and reconstruction methods.

METHODS

²³ Na MRI and ²³ Na inversion recovery (IR) MRI of a phantom and simulated brain with and without lesions of different volumes (V = 1.3-38.2 nominal voxels) were simulated 100 times by adding Gaussian noise matching the SNR of real 3T measurements. Each simulation was reconstructed with four different reconstruction methods (Gridding without and with Hamming filter, Compressed sensing (CS) reconstruction without and with anatomical ¹ H prior information). Based on the mean signals within the lesion volumes of simulations with and without lesions, receiver operating characteristics (ROC) were determined and the area under the curve (AUC) was calculated to assess the classifiability for each lesion volume.

RESULTS

Lesions show higher classifiability in ²³ Na MRI than in ²³ Na IR MRI. For typical parameters and SNR of a 3T scan, the voxel normed minimal classifiable lesion volume (AUC > 0.9) is 2.8 voxels for ²³ Na MRI and 19 voxels for ²³ Na IR MRI, respectively. In terms of classifiability, Gridding with Hamming filter and CS without anatomical ¹ H prior outperform CS reconstruction with anatomical ¹ H prior.

CONCLUSION

Reliability of lesion classifiability strongly depends on the lesion volume and the ²³ Na MRI contrast. Additional incorporation of ¹ H prior information in the CS reconstruction was not beneficial for the classification of small MS-like lesions in ²³ Na MRI.

Breast Tumor Identification in Ultrafast MRI Using Temporal and Spatial Information

Simple Summary

The diagnosis of breast cancer with MRI is based on both morphological evaluation and kinetic curve assessment. Current computer-aided diagnosis methods for malignancy determination mainly focus on morphology features but ignored the temporal information in dynamic contrast-enhanced MRI sequences. Malignant and benign lesions usually have different enhancement patterns during the wash-in phase. Ultrafast breast MRI with high temporal resolution can capture the inflow of contrast in breast lesions. This advantage of ultrafast MRI enables the combination of both temporal and spatial information for automatic breast lesion analysis model development. We found that temporal information helps to significantly improve the performance of breast lesion classification. This suggests that ultrafast MRI provides useful information for malignancy identification and temporal information, which is indispensable for similar model development.

Abstract

Purpose: To investigate the feasibility of using deep learning methods to differentiate benign from malignant breast lesions in ultrafast MRI with both temporal and spatial information. Methods: A total of 173 single breasts of 122 women (151 examinations) with lesions above 5 mm were retrospectively included. A total of 109 out of 173 lesions were benign. Maximum intensity projection (MIP) images were generated from each of the 14 contrast-enhanced T1-weighted acquisitions in the ultrafast MRI scan. A 2D convolutional neural network (CNN) and a long short-term memory (LSTM) network were employed to extract morphological and temporal features, respectively. The 2D CNN model was trained with the MIPs from the last four acquisitions to ensure the visibility of the lesions, while the LSTM model took MIPs of an entire scan as input. The performance of each model and their combination were evaluated with 100-times repeated stratified four-fold cross-validation. Those models were then compared with models developed with standard DCE-MRI which followed the same data split. Results: In the differentiation between benign and malignant lesions, the ultrafast MRI-based 2D CNN achieved a mean AUC of 0.81 ± 0.06, and the LSTM network achieved a mean AUC of 0.78 ± 0.07; their combination showed a mean AUC of 0.83 ± 0.06 in the cross-validation. The mean AUC values were significantly higher for ultrafast MRI-based models than standard DCE-MRI-based models. Conclusion: Deep learning models developed with ultrafast breast MRI achieved higher performances than standard DCE-MRI for malignancy discrimination. The improved AUC values of the combined models indicate an added value of temporal information extracted by the LSTM model in breast lesion characterization.

Evaluation of a pipeline for simulation, reconstruction, and classification in ultrasound-aided diffuse optical tomography of breast tumors

SIGNIFICANCE

Diffuse optical tomography is an ill-posed problem. Combination with ultrasound can improve the results of diffuse optical tomography applied to the diagnosis of breast cancer and allow for classification of lesions.

AIM

To provide a simulation pipeline for the assessment of reconstruction and classification methods for diffuse optical tomography with concurrent ultrasound information.

APPROACH

A set of breast digital phantoms with benign and malignant lesions was simulated building on the software VICTRE. Acoustic and optical properties were assigned to the phantoms for the generation of B-mode images and optical data. A reconstruction algorithm based on a two-region nonlinear fitting and incorporating the ultrasound information was tested. Machine learning classification methods were applied to the reconstructed values to discriminate lesions into benign and malignant after reconstruction.

RESULTS

The approach allowed us to generate realistic US and optical data and to test a two-region reconstruction method for a large number of realistic simulations. When information is extracted from ultrasound images, at least 75% of lesions are correctly classified. With ideal two-region separation, the accuracy is higher than 80%.

CONCLUSIONS

A pipeline for the generation of realistic ultrasound and diffuse optics data was implemented. Machine learning methods applied to a optical reconstruction with a nonlinear optical model and morphological information permit to discriminate malignant lesions from benign ones.

3D Virtual Pancreatography

We present 3D virtual pancreatography (VP), a novel visualization procedure and application for non-invasive diagnosis and classification of pancreatic lesions, the precursors of pancreatic cancer. Currently, non-invasive screening of patients is performed through visual inspection of 2D axis-aligned CT images, though the relevant features are often not clearly visible nor automatically detected. VP is an end-to-end visual diagnosis system that includes: A machine learning based automatic segmentation of the pancreatic gland and the lesions, a semi-automatic approach to extract the primary pancreatic duct, a machine learning based automatic classification of lesions into four prominent types, and specialized 3D and 2D exploratory visualizations of the pancreas, lesions and surrounding anatomy. We combine volume rendering with pancreas- and lesion-centric visualizations and measurements for effective diagnosis. We designed VP through close collaboration and feedback from expert radiologists, and evaluated it on multiple real-world CT datasets with various pancreatic lesions and case studies examined by the expert radiologists.

AutoProstate: Towards Automated Reporting of Prostate MRI for Prostate Cancer Assessment Using Deep Learning

Simple Summary

International guidelines recommend multiparametric magnetic resonance imaging (mpMRI) of the prostate for use by radiologists to identify lesions containing clinically significant prostate cancer, prior to confirmatory biopsy. Automatic assessment of prostate mpMRI using artificial intelligence algorithms holds a currently unrealized potential to improve the diagnostic accuracy achievable by radiologists alone, improve the reporting consistency between radiologists, and enhance reporting quality. In this work, we introduce AutoProstate: a deep learning-powered framework for automatic MRI-based prostate cancer assessment. In particular, AutoProstate utilizes patient data and biparametric MRI to populate an automatic web-based report which includes segmentations of the whole prostate, prostatic zones, and candidate clinically significant prostate cancer lesions, and in addition, several derived characteristics with clinical value are presented. Notably, AutoProstate performed well in external validation using the PICTURE study dataset, suggesting value in prospective multicentre validation, with a view towards future deployment into the prostate cancer diagnostic pathway.

Abstract

Multiparametric magnetic resonance imaging (mpMRI) of the prostate is used by radiologists to identify, score, and stage abnormalities that may correspond to clinically significant prostate cancer (CSPCa). Automatic assessment of prostate mpMRI using artificial intelligence algorithms may facilitate a reduction in missed cancers and unnecessary biopsies, an increase in inter-observer agreement between radiologists, and an improvement in reporting quality. In this work, we introduce AutoProstate, a deep learning-powered framework for automatic MRI-based prostate cancer assessment. AutoProstate comprises of three modules: Zone-Segmenter, CSPCa-Segmenter, and Report-Generator. Zone-Segmenter segments the prostatic zones on T2-weighted imaging, CSPCa-Segmenter detects and segments CSPCa lesions using biparametric MRI, and Report-Generator generates an automatic web-based report containing four sections: Patient Details, Prostate Size and PSA Density, Clinically Significant Lesion Candidates, and Findings Summary. In our experiment, AutoProstate was trained using the publicly available PROSTATEx dataset, and externally validated using the PICTURE dataset. Moreover, the performance of AutoProstate was compared to the performance of an experienced radiologist who prospectively read PICTURE dataset cases. In comparison to the radiologist, AutoProstate showed statistically significant improvements in prostate volume and prostate-specific antigen density estimation. Furthermore, AutoProstate matched the CSPCa lesion detection sensitivity of the radiologist, which is paramount, but produced more false positive detections.

Improving Colonoscopy Lesion Classification Using Semi-Supervised Deep Learning

While data-driven approaches excel at many image analysis tasks, the performance of these approaches is often limited by a shortage of annotated data available for training. Recent work in semi-supervised learning has shown that meaningful representations of images can be obtained from training with large quantities of unlabeled data, and that these representations can improve the performance of supervised tasks. Here, we demonstrate that an unsupervised jigsaw learning task, in combination with supervised training, results in up to a 9.8% improvement in correctly classifying lesions in colonoscopy images when compared to a fully-supervised baseline. We additionally benchmark improvements in domain adaptation and out-of-distribution detection, and demonstrate that semi-supervised learning outperforms supervised learning in both cases. In colonoscopy applications, these metrics are important given the skill required for endoscopic assessment of lesions, the wide variety of endoscopy systems in use, and the homogeneity that is typical of labeled datasets.

Microstructural and metabolic changes in the longitudinal progression of white matter hyperintensities

Our purpose is to evaluate the microstructural and metabolism property in the white matter that later become white matter hyperintensity (WMH), and of WMH that later disappeared. Forty subjects with two-year follow-up were included. Each subject had 3DT1, T2FLAIR, DTI and FDG-PET scans. White matter was classified into: constant WMH, growing WMH, shrinking WMH and normal appearing white matter (NAWM). The average DTI (FA and MD) and FDG-PET (standardized FDG-PET rSUV) of each of the above-mentioned region were extracted and compared. At baseline, the growing WMH had lower FA and FDG-PET rSUV than NAWM, but had higher FA than the constant WMH. Longitudinally, in NAWM, there was a more rapid decline in metabolism compared to WMH areas, while in the growing WMH, a progression in diffusion was found. Finally, we discovered that the shrinking WMH had a similar microstructural and metabolism property and progression to the constant WMH. Our results suggest there are dynamic changes in microstructural and metabolism in WMH. The metabolic change was mainly found in NAWM, while the microstructural change was mainly found in WMH region. Besides, the reduced volume in WMH, to a larger extent, is irrelevant to the microstructural or metabolism recovery.

Breast Cancer Detection and Diagnosis Using Mammographic Data: Systematic Review

BACKGROUND

Machine learning (ML) has become a vital part of medical imaging research. ML methods have evolved over the years from manual seeded inputs to automatic initializations. The advancements in the field of ML have led to more intelligent and self-reliant computer-aided diagnosis (CAD) systems, as the learning ability of ML methods has been constantly improving. More and more automated methods are emerging with deep feature learning and representations. Recent advancements of ML with deeper and extensive representation approaches, commonly known as deep learning (DL) approaches, have made a very significant impact on improving the diagnostics capabilities of the CAD systems.

OBJECTIVE

This review aimed to survey both traditional ML and DL literature with particular application for breast cancer diagnosis. The review also provided a brief insight into some well-known DL networks.

METHODS

In this paper, we present an overview of ML and DL techniques with particular application for breast cancer. Specifically, we search the PubMed, Google Scholar, MEDLINE, ScienceDirect, Springer, and Web of Science databases and retrieve the studies in DL for the past 5 years that have used multiview mammogram datasets.

RESULTS

The analysis of traditional ML reveals the limited usage of the methods, whereas the DL methods have great potential for implementation in clinical analysis and improve the diagnostic capability of existing CAD systems.

CONCLUSIONS

From the literature, it can be found that heterogeneous breast densities make masses more challenging to detect and classify compared with calcifications. The traditional ML methods present confined approaches limited to either particular density type or datasets. Although the DL methods show promising improvements in breast cancer diagnosis, there are still issues of data scarcity and computational cost, which have been overcome to a significant extent by applying data augmentation and improved computational power of DL algorithms.

PROSTATEx Challenges for computerized classification of prostate lesions from multiparametric magnetic resonance images

Grand challenges stimulate advances within the medical imaging research community; within a competitive yet friendly environment, they allow for a direct comparison of algorithms through a well-defined, centralized infrastructure. The tasks of the two-part PROSTATEx Challenges (the PROSTATEx Challenge and the PROSTATEx-2 Challenge) are (1) the computerized classification of clinically significant prostate lesions and (2) the computerized determination of Gleason Grade Group in prostate cancer, both based on multiparametric magnetic resonance images. The challenges incorporate well-vetted cases for training and testing, a centralized performance assessment process to evaluate results, and an established infrastructure for case dissemination, communication, and result submission. In the PROSTATEx Challenge, 32 groups apply their computerized methods (71 methods total) to 208 prostate lesions in the test set. The area under the receiver operating characteristic curve for these methods in the task of differentiating between lesions that are and are not clinically significant ranged from 0.45 to 0.87; statistically significant differences in performance among the top-performing methods, however, are not observed. In the PROSTATEx-2 Challenge, 21 groups apply their computerized methods (43 methods total) to 70 prostate lesions in the test set. When compared with the reference standard, the quadratic-weighted kappa values for these methods in the task of assigning a five-point Gleason Grade Group to each lesion range from - 0.24 to 0.27; superiority to random guessing can be established for only two methods. When approached with a sense of commitment and scientific rigor, challenges foster interest in the designated task and encourage innovation in the field.

Skin Cancer Classification Using Convolutional Neural Networks: Systematic Review

Background

State-of-the-art classifiers based on convolutional neural networks (CNNs) were shown to classify images of skin cancer on par with dermatologists and could enable lifesaving and fast diagnoses, even outside the hospital via installation of apps on mobile devices. To our knowledge, at present there is no review of the current work in this research area.

Objective

This study presents the first systematic review of the state-of-the-art research on classifying skin lesions with CNNs. We limit our review to skin lesion classifiers. In particular, methods that apply a CNN only for segmentation or for the classification of dermoscopic patterns are not considered here. Furthermore, this study discusses why the comparability of the presented procedures is very difficult and which challenges must be addressed in the future.

Methods

We searched the Google Scholar, PubMed, Medline, ScienceDirect, and Web of Science databases for systematic reviews and original research articles published in English. Only papers that reported sufficient scientific proceedings are included in this review.

Results

We found 13 papers that classified skin lesions using CNNs. In principle, classification methods can be differentiated according to three principles. Approaches that use a CNN already trained by means of another large dataset and then optimize its parameters to the classification of skin lesions are the most common ones used and they display the best performance with the currently available limited datasets.

Conclusions

CNNs display a high performance as state-of-the-art skin lesion classifiers. Unfortunately, it is difficult to compare different classification methods because some approaches use nonpublic datasets for training and/or testing, thereby making reproducibility difficult. Future publications should use publicly available benchmarks and fully disclose methods used for training to allow comparability.

Correcting partial volume effects in biexponential T2 estimation of small lesions

PURPOSE

T2 mapping provides a quantitative approach for focal liver lesion characterization. For small lesions, a biexponential model should be used to account for partial volume effects (PVE). However, conventional biexponential fitting suffers from large uncertainty of the fitted parameters when noise is present. The purpose of this work is to develop a more robust method to correct for PVE affecting small lesions.

METHODS

We developed a region of interest-based joint biexponential fitting (JBF) algorithm to estimate the T2 of lesions affected by PVE. JBF takes advantage of the lesion fraction variation among voxels within a region of interest. JBF is compared to conventional approaches using Cramér-Rao lower bound analysis, numerical simulations, phantom, and in vivo data.

RESULTS

JBF provides more accurate and precise T2 estimates in the presence of PVE. Furthermore, JBF is less sensitive to region of interest drawing. Phantom and in vivo results show that JBF can be combined with a reconstruction method for highly undersampled data, enabling the characterization of small abdominal lesions from data acquired in a single breath hold.

CONCLUSION

The JBF algorithm provides more accurate and stable T2 estimates for small structures than conventional techniques when PVE is present. It should be particularly useful for the characterization of small abdominal lesions.

Classification of lesion area in stroke patients during the subacute phase: a multiparametric MRI study

PURPOSE

Stroke imaging studies during the acute phase are likely to precede several vascular brain mechanisms, which have an important role in patient outcome. The aim of this study was to identify within the lesion area during the subacute phase (≥1 day) reactive tissue, which may have the potential for recovery.

METHODS

Twenty seven stroke patients from two cohorts were included. MRI performed during the subacute phase included conventional, perfusion and diffusion imaging. In cohort I, unsupervised multiparametric classification of the lesion area was performed. In cohort II threshold based classification was performed during the subacute phase, and radiological outcome was assessed at follow-up scan.

RESULTS

Three tissue classes were identified in cohort I, referred to as irreversibly damaged, intermediary, and reactive tissue. Based on threshold values defined in cohort I, the reactive tissue was identified in 11/13 patients in cohort II, and showed tissue preservation/partial recovery in 9/11 patients at follow-up scan. The irreversibly damaged tissue was identified in 7/13 patients in cohort II, and predicted tissue necrosis in all cases.

CONCLUSION

Identification of reactive tissue following stroke during the subacute phase can improve radiological assessment, contribute to the understanding of brain recovery processes and has implications for new therapeutic approaches.

Classification of breast lesions pre-contrast injection using water resonance lineshape analysis

Inhomogeneously broadened, non-Lorentzian water resonances have been observed in small image voxels of breast tissue. The non-Lorentzian components of the water resonance are probably produced by bulk magnetic susceptibility shifts caused by dense, deoxygenated tumor blood vessels (the 'blood oxygenation level-dependent' effect), but can also be produced by other characteristics of local anatomy and physiology, including calcifications and interfaces between different types of tissue. Here, we tested the hypothesis that the detection of non-Lorentzian components of the water resonance with high spectral and spatial resolution (HiSS) MRI allows the classification of breast lesions without the need to inject contrast agent. Eighteen malignant lesions and nine benign lesions were imaged with HiSS MRI at 1.5 T. A new algorithm was developed to detect non-Lorentzian (or off-peak) components of the water resonance. After a Lorentzian fit had been subtracted from the data, the largest peak in the residual spectrum in each voxel was identified as the major off-peak component of the water resonance. The difference in frequency between these off-peak components and the main water peaks, and their amplitudes, were measured in malignant lesions, benign lesions and breast fibroglandular tissue. Off-peak component frequencies were significantly different between malignant and benign lesions (p < 0.001). Receiver operating characteristic (ROC) analysis was used to assess the diagnostic performance of HiSS off-peak component analysis compared with dynamic contrast-enhanced (DCE) MRI parameters. The areas under the ROC curves for the 'DCE rapid uptake fraction', 'DCE washout fraction', 'off-peak component amplitude' and 'off-peak component frequency' were 0.75, 0.83, 0.50 and 0.86, respectively. These results suggest that water resonance lineshape analysis performs well in the classification of breast lesions without contrast injection and could improve the diagnostic accuracy of clinical breast MR examinations. In addition, this approach may provide an alternative to DCE MRI in women who are at risk for adverse reactions to contrast media.