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

Ensemble Deep Learning-Based Image Classification for Breast Cancer Subtype and Invasiveness Diagnosis from Whole Slide Image Histopathology

Cancer diagnosis and classification are pivotal for effective patient management and treatment planning. In this study, a comprehensive approach is presented utilizing ensemble deep learning techniques to analyze breast cancer histopathology images. Our datasets were based on two widely employed datasets from different centers for two different tasks: BACH and BreakHis. Within the BACH dataset, a proposed ensemble strategy was employed, incorporating VGG16 and ResNet50 architectures to achieve precise classification of breast cancer histopathology images. Introducing a novel image patching technique to preprocess a high-resolution image facilitated a focused analysis of localized regions of interest. The annotated BACH dataset encompassed 400 WSIs across four distinct classes: Normal, Benign, In Situ Carcinoma, and Invasive Carcinoma. In addition, the proposed ensemble was used on the BreakHis dataset, utilizing VGG16, ResNet34, and ResNet50 models to classify microscopic images into eight distinct categories (four benign and four malignant). For both datasets, a five-fold cross-validation approach was employed for rigorous training and testing. Preliminary experimental results indicated a patch classification accuracy of 95.31% (for the BACH dataset) and WSI image classification accuracy of 98.43% (BreakHis). This research significantly contributes to ongoing endeavors in harnessing artificial intelligence to advance breast cancer diagnosis, potentially fostering improved patient outcomes and alleviating healthcare burdens.

Association of Total Hip Arthroplasty Flexural Rigidity With Magnetic Resonance Imaging and Histological Findings

BACKGROUND

Modular connections in total hip arthroplasty (THA) offer surgical advantages, but can contribute to implant fretting and corrosion due to micromotion at the head-stem interface. Previous studies implicated lower flexural rigidity as a key contributing factor to THA corrosion and fretting, but none associated flexural rigidity with direct histological evaluation or magnetic resonance imaging (MRI) outcomes. The purpose of this study was to determine how implant flexural rigidity is associated with MRI imaging metrics and histopathological outcomes in patients who have a failed THA.

METHODS

Patients requiring revision THA surgery underwent preoperative MRIs with 3-dimensional multispectral imaging techniques to suppress metal artifacts. The MRI images were graded for adverse local tissue reactions. For each hip, trunnion flexural rigidity was measured from the retrieved femoral stem, and a periprosthetic tissue sample was retrieved and evaluated using semiquantitative histology. Generalized linear models and analyses of variance were used to assess associations between flexural rigidity and MRI and histology outcomes.

RESULTS

A total of 106 THA stems were retrieved (46 women and 60 men, age: 68 years (range, 60 to 73 years). After adjustment for length of implantation, flexural rigidity was negatively correlated with histologic aseptic lymphocyte-dominant vasculitis-associated lesion severity (β = -26.27, P = .018), Fujishiro lymphocyte grading (β = -13.4, P = .039), perivascular lymphocyte layers (β = -17.8, P = .022), the grade of tissue organization (β = -22.5, P = .009), the presence of diffuse synovitis (β = -66.5, P = .003), and the presence of lymphoid aggregates (β = -75.9, P = .022). No association was found between MRI metrics and flexural rigidity.

CONCLUSIONS

Among these implants, decreased trunnion stiffness was associated with increased histologic features of adverse host-mediated soft tissue reactions.

Multi-model sequential analysis of MRI data for microstructure prediction in heterogeneous tissue

We propose a general method for combining multiple models to predict tissue microstructure, with an exemplar using in vivo diffusion-relaxation MRI data. The proposed method obviates the need to select a single 'optimum' structure model for data analysis in heterogeneous tissues where the best model varies according to local environment. We break signal interpretation into a three-stage sequence: (1) application of multiple semi-phenomenological models to predict the physical properties of tissue water pools contributing to the observed signal; (2) from each Stage-1 semi-phenomenological model, application of a tissue microstructure model to predict the relative volumes of tissue structure components that make up each water pool; and (3) aggregation of the predictions of tissue structure, with weightings based on model likelihood and fractional volumes of the water pools from Stage-1. The multiple model approach is expected to reduce prediction variance in tissue regions where a complex model is overparameterised, and bias where a model is underparameterised. The separation of signal characterisation (Stage-1) from biological assignment (Stage-2) enables alternative biological interpretations of the observed physical properties of the system, by application of different tissue structure models. The proposed method is exemplified with human prostate diffusion-relaxation MRI data, but has potential application to a wide range of analyses where a single model may not be optimal throughout the sampled domain.

Recent Progress in Printed Photonic Devices: A Brief Review of Materials, Devices, and Applications

Printing electronics incorporates several significant technologies, such as semiconductor devices produced by various printing techniques on flexible substrates. With the growing interest in printed electronic devices, new technologies have been developed to make novel devices with inexpensive and large-area printing techniques. This review article focuses on the most recent developments in printed photonic devices. Photonics and optoelectronic systems may now be built utilizing materials with specific optical properties and 3D designs achieved through additive printing. Optical and architected materials that can be printed in their entirety are among the most promising future research topics, as are platforms for multi-material processing and printing technologies that can print enormous volumes at a high resolution while also maintaining a high throughput. Significant advances in innovative printable materials create new opportunities for functional devices to act efficiently, such as wearable sensors, integrated optoelectronics, and consumer electronics. This article provides an overview of printable materials, printing methods, and the uses of printed electronic devices.

Developments in functional imaging of the placenta

The placenta is both the literal and metaphorical black box of pregnancy. Measurement of the function of the placenta has the potential to enhance our understanding of this enigmatic organ and serve to support obstetric decision making. Advanced imaging techniques are key to support these measurements. This review summarises emerging imaging technology being used to measure the function of the placenta and new developments in the computational analysis of these data. We address three important examples where functional imaging is supporting our understanding of these conditions: fetal growth restriction, placenta accreta, and twin-twin transfusion syndrome.

Temperature correction of post mortem quantitative magnetic resonance imaging using real-time forehead temperature acquisitions

Performing magnetic resonance imaging (MRI) of deceased is challenging due to altered body temperatures compared to in vivo temperatures and, hence, requires a temperature correction. This study investigates the possibility to correct brain MRI parameters real-time and non invasively based on the forehead temperature. 17 post mortem cases were included and their forehead temperatures were measured continuously during the in situ brain MRI protocol consisting of a diffusion tensor imaging, multi-contrast spin echo, multi-echo gradient echo and inversion recovery spin echo sequence. Linear models were fitted to the quantitative MRI parameters in a forensically interesting temperature range for white matter, cerebral cortex and deep gray matter, separately, and the influence of the forehead temperature on the MRI parameters was determined. A statistically significant temperature sensitivity was found for T₂ and mean diffusivity in white matter, for T₁ in cerebral cortex, as well as for T₁ and mean diffusivity in deep gray matter. Linear models were computed to temperature correct these MRI parameters in in situ post mortem scans to allow their comparison regardless of temperature. The here presented real-time and non invasive temperature correction method for the brain presents a crucial precondition for quantitative in situ post mortem MRI.

Integrating multimodality magnetic resonance imaging to the Allen Mouse Brain Common Coordinate Framework

High-resolution magnetic resonance imaging (MRI) affords unique image contrasts to nondestructively probe the tissue microstructure; validation of MRI findings with conventional histology is essential to better understand the MRI contrasts. However, the dramatic difference in the spatial resolution and image contrast of these two techniques impedes accurate comparison between MRI metrics and traditional histology. To better validate various MRI metrics, we acquired whole mouse brain multigradient recalled-echo and multishell diffusion MRI datasets at 25-μm isotropic resolution. The recently developed Allen Mouse Brain Common Coordinate Framework (CCFv3) provides opportunities to integrate multimodal and multiscale datasets of the whole mouse brain in a common three-dimensional (3D) space. The T2*, quantitative susceptibility mapping, diffusion tensor imaging, and neurite orientation dispersion and density imaging parameters were compared with both serial two-photon tomography images and 3D Nissl staining images in the CCFv3 at the same spatial resolution. The correlation between MRI and Nissl staining strongly depends on different metrics and different regions of the brain. Integrating different imaging modalities to the same space may substantially improve our understanding of the complexity of the brain at different scales.

Radiologic-pathologic correlation of lesions in resected liver specimens with an ex vivo MRI-compatible localization device

OBJECTIVE

Liver lesion characterization is limited by the lack of an established gold standard for precise correlation of radiologic characteristics with their histologic features. The objective of this study was to demonstrate the feasibility of using an ex vivo MRI-compatible sectioning device for radiologic-pathologic co-localization of lesions in resected liver specimens.

METHODS

In this prospective feasibility study, adults undergoing curative partial hepatectomy from February 2018 to January 2019 were enrolled. Gadoxetic acid was administered intraoperatively prior to hepatic vascular inflow ligation. Liver specimens were stabilized in an MRI-compatible acrylic lesion localization device (27 × 14 × 14 cm3) featuring slicing channels and a silicone gel 3D matrix. High-resolution 3D T1-weighted fast spoiled gradient echo and 3D T2-weighted fast-spin-echo images were acquired using a single channel quadrature head coil. Radiologic lesion coordinates guided pathologic sectioning. A final histopathologic diagnosis was prepared for all lesions. The proportion of successfully co-localized lesions was determined.

RESULTS

A total of 57 lesions were identified radiologically and sectioned in liver specimens from 10 participants with liver metastases (n = 8), primary biliary mucinous cystic neoplasm (n = 1), and hepatic adenomatosis (n = 1). Of these, 38 lesions (67%) were < 1 cm. Overall, 52/57 (91%) of radiologically identified lesions were identified pathologically using the device. Of these, 5 lesions (10%) were not initially identified on gross examination but were confirmed histologically using MRI-guided localization. One lesion was identified grossly but not on MRI.

CONCLUSIONS

We successfully demonstrated the feasibility of a clinical method for image-guided co-localization and histological characterization of liver lesions using an ex vivo MRI-compatible sectioning device.

KEY POINTS

• The ex vivo MRI-compatible sectioning device provides a reliable method for radiologic-pathologic correlation of small (< 1 cm) liver lesions in human liver specimens. • The sectioning method can be feasibly implemented within a clinical practice setting and used in future efforts to study liver lesion characterization. • Intraoperative administration of gadoxetic acid results in enhancement in ex vivo MRI images of liver specimens hours later with excellent image quality.

Tensor image registration library: Deformable registration of stand-alone histology images to whole-brain post-mortem MRI data

BACKGROUND

Accurate registration between microscopy and MRI data is necessary for validating imaging biomarkers against neuropathology, and to disentangle complex signal dependencies in microstructural MRI. Existing registration methods often rely on serial histological sampling or significant manual input, providing limited scope to work with a large number of stand-alone histology sections. Here we present a customisable pipeline to assist the registration of stand-alone histology sections to whole-brain MRI data.

METHODS

Our pipeline registers stained histology sections to whole-brain post-mortem MRI in 4 stages, with the help of two photographic intermediaries: a block face image (to undistort histology sections) and coronal brain slab photographs (to insert them into MRI space). Each registration stage is implemented as a configurable stand-alone Python script using our novel platform, Tensor Image Registration Library (TIRL), which provides flexibility for wider adaptation. We report our experience of registering 87 PLP-stained histology sections from 14 subjects and perform various experiments to assess the accuracy and robustness of each stage of the pipeline.

RESULTS

All 87 histology sections were successfully registered to MRI. Histology-to-block registration (Stage 1) achieved 0.2-0.4 mm accuracy, better than commonly used existing methods. Block-to-slice matching (Stage 2) showed great robustness in automatically identifying and inserting small tissue blocks into whole brain slices with 0.2 mm accuracy. Simulations demonstrated sub-voxel level accuracy (0.13 mm) of the slice-to-volume registration (Stage 3) algorithm, which was observed in over 200 actual brain slice registrations, compensating 3D slice deformations up to 6.5 mm. Stage 4 combined the previous stages and generated refined pixelwise aligned multi-modal histology-MRI stacks.

CONCLUSIONS

Our open-source pipeline provides robust automation tools for registering stand-alone histology sections to MRI data with sub-voxel level precision, and the underlying framework makes it readily adaptable to a diverse range of microscopy-MRI studies.

Technique to match mesorectal lymph nodes imaging findings to histopathology: node-by-node comparison

BACKGROUND

Lymph node status is critical for staging rectal cancer and determining neoadjuvant therapy regimens. Establishing a matching between imaging and histopathological lymph nodes is fundamental for predicting lymph node status. This study reports a technique to achieve node-by-node pairing of mesorectal lymph nodes between imaging findings and histopathology.

METHODS

Fifty-two patients with histopathologically verified rectal cancer underwent magnetic resonance imaging before surgery. The status of each lymph node in the surgical specimens was analyzed histopathologically and matched with preoperative imaging after the operation.

RESULTS

A total of 346 mesorectal lymph nodes were located on imaging evaluation, of which 313 were confirmed histopathologically, and 33 were unmatched. The total success rate of the technique was 90.5%. Node-by-node analysis revealed 280 benign and 33 malignant nodal structures.

CONCLUSION

The technique to match mesorectal lymph node imaging findings to histopathology was feasible and effective. It simplified the technical method and had a reasonable success matching rate, which could provide a standardized approach for obtaining a prospective correlation between imaging and histological findings, supporting all subsequent related studies at the level of mesorectal lymph nodes.

Post mortem brain temperature and its influence on quantitative MRI of the brain

Objective

MRI temperature sensitivity presents a major issue in in situ post mortem MRI (PMMRI), as the tissue temperatures differ from living persons due to passive cooling of the deceased. This study aims at computing brain temperature effects on the MRI parameters to correct for temperature in PMMRI, laying the foundation for future projects on post mortem validation of in vivo MRI techniques.

Materials and methods

Brain MRI parameters were assessed in vivo and in situ post mortem using a 3 T MRI scanner. Post mortem brain temperature was measured in situ transethmoidally. The temperature effect was computed by fitting a linear model to the MRI parameters and the corresponding brain temperature.

Results

Linear positive temperature correlations were observed for T₁, T₂* and mean diffusivity in all tissue types. A significant negative correlation was observed for T₂ in white matter. Fractional anisotropy revealed significant correlations in all gray matter regions except for the thalamus.

Discussion

The linear models will allow to correct for temperature in post mortem MRI. Comparing in vivo to post mortem conditions, the mean diffusivity, in contrast to T₁ and T₂, revealed additional effects besides temperature, such as cessation of perfusion and active diffusion.

Elastic transformation of histological slices allows precise co-registration with microCT data sets for a refined virtual histology approach

Although X-ray based 3D virtual histology is an emerging tool for the analysis of biological tissue, it falls short in terms of specificity when compared to conventional histology. Thus, the aim was to establish a novel approach that combines 3D information provided by microCT with high specificity that only (immuno-)histochemistry can offer. For this purpose, we developed a software frontend, which utilises an elastic transformation technique to accurately co-register various histological and immunohistochemical stainings with free propagation phase contrast synchrotron radiation microCT. We demonstrate that the precision of the overlay of both imaging modalities is significantly improved by performing our elastic registration workflow, as evidenced by calculation of the displacement index. To illustrate the need for an elastic co-registration approach we examined specimens from a mouse model of breast cancer with injected metal-based nanoparticles. Using the elastic transformation pipeline, we were able to co-localise the nanoparticles to specifically stained cells or tissue structures into their three-dimensional anatomical context. Additionally, we performed a semi-automated tissue structure and cell classification. This workflow provides new insights on histopathological analysis by combining CT specific three-dimensional information with cell/tissue specific information provided by classical histology.