Systems genetics uncover new loci containing functional gene candidates in Mycobacterium tuberculosis-infected Diversity Outbred mice.

Mycobacterium tuberculosis infects two billion people across the globe, and results in 8-9 million new tuberculosis (TB) cases and 1-1.5 million deaths each year. Most patients have no known genetic basis that predisposes them to disease. Here, we investigate the complex genetic basis of pulmonary TB by modelling human genetic diversity with the Diversity Outbred mouse population. When infected with M. tuberculosis, one-third develop early onset, rapidly progressive, necrotizing granulomas and succumb within 60 days. The remaining develop non-necrotizing granulomas and survive longer than 60 days. Genetic mapping using immune and inflammatory mediators; and clinical, microbiological, and granuloma correlates of disease identified five new loci on mouse chromosomes 1, 2, 4, 16; and three known loci on chromosomes 3 and 17. Further, multiple positively correlated traits shared loci on chromosomes 1, 16, and 17 and had similar patterns of allele effects, suggesting these loci contain critical genetic regulators of inflammatory responses to M. tuberculosis. To narrow the list of candidate genes, we used a machine learning strategy that integrated gene expression signatures from lungs of M. tuberculosis-infected Diversity Outbred mice with gene interaction networks to generate scores representing functional relationships. The scores were used to rank candidates for each mapped trait, resulting in 11 candidate genes: Ncf2, Fam20b, S100a8, S100a9, Itgb5, Fstl1, Zbtb20, Ddr1, Ier3, Vegfa, and Zfp318. Although all candidates have roles in infection, inflammation, cell migration, extracellular matrix remodeling, or intracellular signaling, and all contain single nucleotide polymorphisms (SNPs), SNPs in only four genes (S100a8, Itgb5, Fstl1, Zfp318) are predicted to have deleterious effects on protein functions. We performed methodological and candidate validations to (i) assess biological relevance of predicted allele effects by showing that Diversity Outbred mice carrying PWK/PhJ alleles at the H-2 locus on chromosome 17 QTL have shorter survival; (ii) confirm accuracy of predicted allele effects by quantifying S100A8 protein in inbred founder strains; and (iii) infection of C57BL/6 mice deficient for the S100a8 gene. Overall, this body of work demonstrates that systems genetics using Diversity Outbred mice can identify new (and known) QTLs and functionally relevant gene candidates that may be major regulators of complex host-pathogens interactions contributing to granuloma necrosis and acute inflammation in pulmonary TB.

One label is all you need: Interpretable AI-enhanced histopathology for oncology.

Artificial Intelligence (AI)-enhanced histopathology presents unprecedented opportunities to benefit oncology through interpretable methods that require only one overall label per hematoxylin and eosin (H&E) slide with no tissue-level annotations. We present a structured review of these methods organized by their degree of verifiability and by commonly recurring application areas in oncological characterization. First, we discuss morphological markers (tumor presence/absence, metastases, subtypes, grades) in which AI-identified regions of interest (ROIs) within whole slide images (WSIs) verifiably overlap with pathologist-identified ROIs. Second, we discuss molecular markers (gene expression, molecular subtyping) that are not verified via H&E but rather based on overlap with positive regions on adjacent tissue. Third, we discuss genetic markers (mutations, mutational burden, microsatellite instability, chromosomal instability) that current technologies cannot verify if AI methods spatially resolve specific genetic alterations. Fourth, we discuss the direct prediction of survival to which AI-identified histopathological features quantitatively correlate but are nonetheless not mechanistically verifiable. Finally, we discuss in detail several opportunities and challenges for these one-label-per-slide methods within oncology. Opportunities include reducing the cost of research and clinical care, reducing the workload of clinicians, personalized medicine, and unlocking the full potential of histopathology through new imaging-based biomarkers. Current challenges include explainability and interpretability, validation via adjacent tissue sections, reproducibility, data availability, computational needs, data requirements, domain adaptability, external validation, dataset imbalances, and finally commercialization and clinical potential. Ultimately, the relative ease and minimum upfront cost with which relevant data can be collected in addition to the plethora of available AI methods for outcome-driven analysis will surmount these current limitations and achieve the innumerable opportunities associated with AI-driven histopathology for the benefit of oncology.

Predicting response to neoadjuvant chemotherapy for colorectal liver metastasis using deep learning on prechemotherapy cross-sectional imaging.

BACKGROUND AND OBJECTIVES: Deep learning models (DLMs) are applied across domains of health sciences to generate meaningful predictions. DLMs make use of neural networks to generate predictions from discrete data inputs. This study employs DLM on prechemotherapy cross-sectional imaging to predict patients' response to neoadjuvant chemotherapy. METHODS: Adult patients with colorectal liver metastasis who underwent surgery after neoadjuvant chemotherapy were included. A DLM was trained on computed tomography images using attention-based multiple-instance learning. A logistic regression model incorporating clinical parameters of the Fong clinical risk score was used for comparison. Both model performances were benchmarked against the Response Evaluation Criteria in Solid Tumors criteria. A receiver operating curve was created and resulting area under the curve (AUC) was determined. RESULTS: Ninety-five patients were included, with 33,619 images available for study inclusion. Ninety-five percent of patients underwent 5-fluorouracil-based chemotherapy with oxaliplatin and/or irinotecan. Sixty percent of the patients were categorized as chemotherapy responders (30% reduction in tumor diameter). The DLM had an AUC of 0.77. The AUC for the clinical model was 0.41. CONCLUSIONS: Image-based DLM for prediction of response to neoadjuvant chemotherapy in patients with colorectal cancer liver metastases was superior to a clinical-based model. These results demonstrate potential to identify nonresponders to chemotherapy and guide select patients toward earlier curative resection.

Ki-67 assessment of pancreatic neuroendocrine neoplasms: Systematic review and meta-analysis of manual vs. digital pathology scoring.

Ki-67 assessment is a key step in the diagnosis of neuroendocrine neoplasms (NENs) from all anatomic locations. Several challenges exist related to quantifying the Ki-67 proliferation index due to lack of method standardization and inter-reader variability. The application of digital pathology coupled with machine learning has been shown to be highly accurate and reproducible for the evaluation of Ki-67 in NENs. We systematically reviewed all published studies on the subject of Ki-67 assessment in pancreatic NENs (PanNENs) employing digital image analysis (DIA). The most common advantages of DIA were improvement in the standardization and reliability of Ki-67 evaluation, as well as its speed and practicality, compared to the current gold standard approach of manual counts from captured images, which is cumbersome and time consuming. The main limitations were attributed to higher costs, lack of widespread availability (as of yet), operator qualification and training issues (if it is not done by pathologists), and most importantly, the drawback of image algorithms counting contaminating non-neoplastic cells and other signals like hemosiderin. However, solutions are rapidly developing for all of these challenging issues. A comparative meta-analysis for DIA versus manual counting shows very high concordance (global coefficient of concordance: 0.94, 95% CI: 0.83-0.98) between these two modalities. These findings support the widespread adoption of validated DIA methods for Ki-67 assessment in PanNENs, provided that measures are in place to ensure counting of only tumor cells either by software modifications or education of non-pathologist operators, as well as selection of standard regions of interest for analysis. NENs, being cellular and monotonous neoplasms, are naturally more amenable to Ki-67 assessment. However, lessons of this review may be applicable to other neoplasms where proliferation activity has become an integral part of theranostic evaluation including breast, brain, and hematolymphoid neoplasms.

Ros-NET: A deep convolutional neural network for automatic identification of rosacea lesions.

BACKGROUND: Rosacea is one of the most common cutaneous disorder characterized primarily by facial flushing, erythema, papules, pustules, telangiectases, and nasal swelling. Diagnosis of rosacea is principally done by a physical examination and a consistent patient history. However, qualitative human assessment is often subjective and suffers from a relatively high intra- and inter-observer variability in evaluating patient outcomes. MATERIALS AND METHODS: To overcome these problems, we propose a quantitative and reproducible computer-aided diagnosis system, Ros-NET, which integrates information from different image scales and resolutions in order to identify rosacea lesions. This involves adaption of Inception-ResNet-v2 and ResNet-101 to extract rosacea features from facial images. Additionally, we propose to refine the detection results by means of facial-landmarks-based zones (ie, anthropometric landmarks) as regions of interest (ROI), which focus on typical areas of rosacea occurrence on a face. RESULTS: Using a leave-one-patient-out cross-validation scheme, the weighted average Dice coefficients, in percentages, across all patients (N = 41) with 256 × 256 image patches are 89.8 ± 2.6% and 87.8 ± 2.4% with Inception-ResNet-v2 and ResNet-101, respectively. CONCLUSION: The findings from this study support that pre-trained networks trained via transfer learning can be beneficial in identifying rosacea lesions. Our future work will involve expanding the work to a larger database of cases with varying degrees of disease characteristics.

Digital pathology and artificial intelligence.

In modern clinical practice, digital pathology has a crucial role and is increasingly a technological requirement in the scientific laboratory environment. The advent of whole-slide imaging, availability of faster networks, and cheaper storage solutions has made it easier for pathologists to manage digital slide images and share them for clinical use. In parallel, unprecedented advances in machine learning have enabled the synergy of artificial intelligence and digital pathology, which offers image-based diagnosis possibilities that were once limited only to radiology and cardiology. Integration of digital slides into the pathology workflow, advanced algorithms, and computer-aided diagnostic techniques extend the frontiers of the pathologist's view beyond a microscopic slide and enable true utilisation and integration of knowledge that is beyond human limits and boundaries, and we believe there is clear potential for artificial intelligence breakthroughs in the pathology setting. In this Review, we discuss advancements in digital slide-based image diagnosis for cancer along with some challenges and opportunities for artificial intelligence in digital pathology.

Tumor Budding Detection System in Whole Slide Pathology Images.

Tumor budding is defined as the presence of single tumor cells or small tumor clusters (less than five cells) that 'bud' from the invasive front of the main tumor. Tumor budding (TB) has recently emerged as an important adverse prognostic factor for many different cancer types. In colorectal carcinoma (CRC), tumor budding has been independently associated with lymph node metastasis and poor outcome. Pathologic assessment of tumor budding by light microscopy requires close evaluation of tumor invasive front on intermediate to high power magnification, entailing locating the 'hotspot' of tumor budding, counting all TB in one high power field, and generating a tumor budding score. By automating these time-consuming tasks, computer-assisted image analysis tools can be helpful for daily pathology practice, since tumor budding reporting is now recommended on select cases. In this paper, we report our work on the development of a tumor budding detection system in CRC from whole-slide Cytokeratin AE1/3 images, based on de novo computer algorithm that automates morphometric analysis of tumor budding.

Relationship between the Ki67 index and its area based approximation in breast cancer.

BACKGROUND: The Ki67 Index has been extensively studied as a prognostic biomarker in breast cancer. However, its clinical adoption is largely hampered by the lack of a standardized method to assess Ki67 that limits inter-laboratory reproducibility. It is important to standardize the computation of the Ki67 Index before it can be effectively used in clincial practice. METHOD: In this study, we develop a systematic approach towards standardization of the Ki67 Index. We first create the ground truth consisting of tumor positive and tumor negative nuclei by registering adjacent breast tissue sections stained with Ki67 and H&E. The registration is followed by segmentation of positive and negative nuclei within tumor regions from Ki67 images. The true Ki67 Index is then approximated with a linear model of the area of positive to the total area of tumor nuclei. RESULTS: When tested on 75 images of Ki67 stained breast cancer biopsies, the proposed method resulted in an average root mean square error of 3.34. In comparison, an expert pathologist resulted in an average root mean square error of 9.98 and an existing automated approach produced an average root mean square error of 5.64. CONCLUSIONS: We show that it is possible to approximate the true Ki67 Index accurately without detecting individual nuclei and also statically demonstrate the weaknesses of commonly adopted approaches that use both tumor and non-tumor regions together while compensating for the latter with higher order approximations.

Developing the Quantitative Histopathology Image Ontology (QHIO): A case study using the hot spot detection problem.

Interoperability across data sets is a key challenge for quantitative histopathological imaging. There is a need for an ontology that can support effective merging of pathological image data with associated clinical and demographic data. To foster organized, cross-disciplinary, information-driven collaborations in the pathological imaging field, we propose to develop an ontology to represent imaging data and methods used in pathological imaging and analysis, and call it Quantitative Histopathological Imaging Ontology - QHIO. We apply QHIO to breast cancer hot-spot detection with the goal of enhancing reliability of detection by promoting the sharing of data between image analysts.

A multi-resolution textural approach to diagnostic neuropathology reporting.

We present a computer aided diagnostic workflow focusing on two diagnostic branch points in neuropathology (intraoperative consultation and p53 status in tumor biopsy specimens) by means of texture analysis via discrete wavelet frames decomposition. For intraoperative consultation, our methodology is capable of classifying glioblastoma versus metastatic cancer by extracting textural features from the non-nuclei region of cytologic preparations based on the imaging characteristics of glial processes, which appear as anisotropic thin linear structures. For metastasis, these are homogeneous in appearance, thus suitable and extractable texture features distinguish the two tissue types. Experiments on 53 images (29 glioblastomas and 24 metastases) resulted in average accuracy as high as 89.7 % for glioblastoma, 87.5 % for metastasis and 88.7 % overall. For p53 interpretation, we detect and classify p53 status by classifying staining intensity into strong, moderate, weak and negative sub-classes. We achieved this by developing a novel adaptive thresholding for detection, a two-step rule based on weighted color and intensity for the classification of positively and negatively stained nuclei, followed by texture classification to classify the positively stained nuclei into the strong, moderate and weak intensity sub-classes. Our detection method is able to correctly locate and distinguish the four types of cells, at 85 % average precision and 88 % average sensitivity rate. These classification methods on the other hand recorded 81 % accuracy in classifying the positive and negative cells, and 60 % accuracy in further classifying the positive cells into the three intensity groups, which is comparable with neuropathologists' markings.

Pten in stromal fibroblasts suppresses mammary epithelial tumours.

The tumour stroma is believed to contribute to some of the most malignant characteristics of epithelial tumours. However, signalling between stromal and tumour cells is complex and remains poorly understood. Here we show that the genetic inactivation of Pten in stromal fibroblasts of mouse mammary glands accelerated the initiation, progression and malignant transformation of mammary epithelial tumours. This was associated with the massive remodelling of the extracellular matrix (ECM), innate immune cell infiltration and increased angiogenesis. Loss of Pten in stromal fibroblasts led to increased expression, phosphorylation (T72) and recruitment of Ets2 to target promoters known to be involved in these processes. Remarkably, Ets2 inactivation in Pten stroma-deleted tumours ameliorated disruption of the tumour microenvironment and was sufficient to decrease tumour growth and progression. Global gene expression profiling of mammary stromal cells identified a Pten-specific signature that was highly represented in the tumour stroma of patients with breast cancer. These findings identify the Pten-Ets2 axis as a critical stroma-specific signalling pathway that suppresses mammary epithelial tumours.

Classification of follicular lymphoma: the effect of computer aid on pathologists grading.

BACKGROUND: Follicular lymphoma (FL) is one of the most common lymphoid malignancies in the western world. FL cases are stratified into three histological grades based on the average centroblast count per high power field (HPF). The centroblast count is performed manually by the pathologist using an optical microscope and hematoxylin and eosin (H&E) stained tissue section. Although this is the current clinical practice, it suffers from high inter- and intra-observer variability and is vulnerable to sampling bias. METHODS: In this paper, we present a system, called Follicular Lymphoma Grading System (FLAGS), to assist the pathologist in grading FL cases. We also assess the effect of FLAGS on accuracy of expert and inexperienced readers. FLAGS automatically identifies possible HPFs for examination by analyzing H&E and CD20 stains, before classifying them into low or high risk categories. The pathologist is first asked to review the slides according to the current routine clinical practice, before being presented with FLAGS classification via color-coded map. The accuracy of the readers with and without FLAGS assistance is measured. RESULTS: FLAGS was used by four experts (board-certified hematopathologists) and seven pathology residents on 20 FL slides. Access to FLAGS improved overall reader accuracy with the biggest improvement seen among residents. An average AUC value of 0.75 was observed which generally indicates "acceptable" diagnostic performance. CONCLUSIONS: The results of this study show that FLAGS can be useful in increasing the pathologists' accuracy in grading the tissue. To the best of our knowledge, this study measure, for the first time, the effect of computerized image analysis on pathologists' grading of follicular lymphoma. When fully developed, such systems have the potential to reduce sampling bias by examining an increased proportion of HPFs within follicle regions, as well as to reduce inter- and intra-reader variability.

Detecting and characterizing cellular responses to Mycobacterium tuberculosis from histology slides.

Infection with Mycobacterium tuberculosis (M.tb) results in immune cell recruitment to the lungs, forming macrophage-rich regions (granulomas) and lymphocyte-rich regions (lymphocytic cuffs). The objective of this study was to accurately identify and characterize these regions from hematoxylin and eosin (H&E)-stained tissue slides. The two target regions (granulomas and lymphocytic cuffs) can be identified by their morphological characteristics. Their most differentiating characteristic on H&E slides is cell density. We developed a computational framework, called DeHiDe, to detect and classify high cell-density regions in histology slides. DeHiDe employed a novel internuclei geodesic distance calculation and Dulmange Mendelsohn permutation to detect and classify high cell-density regions. Lung tissue slides of mice experimentally infected with M.tb were stained with H&E and digitized. A total of 21 digital slides were used to develop and train the computational framework. The performance of the framework was evaluated using two main outcome measures: correct detection of potential regions, and correct classification of potential regions into granulomas and lymphocytic cuffs. DeHiDe provided a detection accuracy of 99.39% while it correctly classified 90.87% of the detected regions for the images where the expert pathologist produced the same ground truth during the first and second round of annotations. We showed that DeHiDe could detect high cell-density regions in a heterogeneous cell environment with non-convex tissue shapes.

Histopathological image analysis for centroblasts classification through dimensionality reduction approaches.

We present two novel automated image analysis methods to differentiate centroblast (CB) cells from noncentroblast (non-CB) cells in digital images of H&E-stained tissues of follicular lymphoma. CB cells are often confused by similar looking cells within the tissue, therefore a system to help their classification is necessary. Our methods extract the discriminatory features of cells by approximating the intrinsic dimensionality from the subspace spanned by CB and non-CB cells. In the first method, discriminatory features are approximated with the help of singular value decomposition (SVD), whereas in the second method they are extracted using Laplacian Eigenmaps. Five hundred high-power field images were extracted from 17 slides, which are then used to compose a database of 213 CB and 234 non-CB region of interest images. The recall, precision, and overall accuracy rates of the developed methods were measured and compared with existing classification methods. Moreover, the reproducibility of both classification methods was also examined. The average values of the overall accuracy were 99.22% ± 0.75% and 99.07% ± 1.53% for COB and CLEM, respectively. The experimental results demonstrate that both proposed methods provide better classification accuracy of CB/non-CB in comparison with the state of the art methods.

Perceptual clustering for automatic hotspot detection from Ki-67-stained neuroendocrine tumour images.

Hotspot detection plays a crucial role in grading of neuroendocrine tumours of the digestive system. Hotspots are often detected manually from Ki-67-stained images, a practice which is tedious, irreproducible and error prone. We report a new method to segment Ki-67-positive nuclei from Ki-67-stained slides of neuroendocrine tumours. The method combines minimal graph cuts along with the multistate difference of Gaussians to detect the individual cells from images of Ki-67-stained slides. It, then, automatically defines the composite function, which is used to determine hotspots in neuroendocrine tumour slide images. We combine modified particle swarm optimization with message passing clustering to mimic the thought process of the pathologist during hotspot detection in neuroendocrine tumour slide images. The proposed method was tested on 55 images of size 10 × 5 K and resulted in an accuracy of 94.60%. The developed methodology can also be part of the workflow for other diseases such as breast cancer and glioblastomas.

Enhancing Colorectal Cancer Tumor Bud Detection Using Deep Learning from Routine H&E-Stained Slides.

Tumor budding refers to a cluster of one to four tumor cells located at the tumor-invasive front. While tumor budding is a prognostic factor for colorectal cancer, counting and grading tumor budding are time consuming and not highly reproducible. There could be high inter- and intra-reader disagreement on H&E evaluation. This leads to the noisy training (imperfect ground truth) of deep learning algorithms, resulting in high variability and losing their ability to generalize on unseen datasets. Pan-cytokeratin staining is one of the potential solutions to enhance the agreement, but it is not routinely used to identify tumor buds and can lead to false positives. Therefore, we aim to develop a weakly-supervised deep learning method for tumor bud detection from routine H&E-stained images that does not require strict tissue-level annotations. We also propose Bayesian Multiple Instance Learning (BMIL) that combines multiple annotated regions during the training process to further enhance the generalizability and stability in tumor bud detection. Our dataset consists of 29 colorectal cancer H&E-stained images that contain 115 tumor buds per slide on average. In six-fold cross-validation, our method demonstrated an average precision and recall of 0.94, and 0.86 respectively. These results provide preliminary evidence of the feasibility of our approach in improving the generalizability in tumor budding detection using H&E images while avoiding the need for non-routine immunohistochemical staining methods.

Translating prognostic quantification of c-MYC and BCL2 from tissue microarrays to whole slide images in diffuse large B-cell lymphoma using deep learning.

BACKGROUND: c-MYC and BCL2 positivity are important prognostic factors for diffuse large B-cell lymphoma. However, manual quantification is subject to significant intra- and inter-observer variability. We developed an automated method for quantification in whole-slide images of tissue sections where manual quantification requires evaluating large areas of tissue with possibly heterogeneous staining. We train this method using annotations of tumor positivity in smaller tissue microarray cores where expression and staining are more homogeneous and then translate this model to whole-slide images. METHODS: Our method applies a technique called attention-based multiple instance learning to regress the proportion of c-MYC-positive and BCL2-positive tumor cells from pathologist-scored tissue microarray cores. This technique does not require annotation of individual cell nuclei and is trained instead on core-level annotations of percent tumor positivity. We translate this model to scoring of whole-slide images by tessellating the slide into smaller core-sized tissue regions and calculating an aggregate score. Our method was trained on a public tissue microarray dataset from Stanford and applied to whole-slide images from a geographically diverse multi-center cohort produced by the Lymphoma Epidemiology of Outcomes study. RESULTS: In tissue microarrays, the automated method had Pearson correlations of 0.843 and 0.919 with pathologist scores for c-MYC and BCL2, respectively. When utilizing standard clinical thresholds, the sensitivity/specificity of our method was 0.743 / 0.963 for c-MYC and 0.938 / 0.951 for BCL2. For double-expressors, sensitivity and specificity were 0.720 and 0.974. When translated to the external WSI dataset scored by two pathologists, Pearson correlation was 0.753 & 0.883 for c-MYC and 0.749 & 0.765 for BCL2, and sensitivity/specificity was 0.857/0.991 & 0.706/0.930 for c-MYC, 0.856/0.719 & 0.855/0.690 for BCL2, and 0.890/1.00 & 0.598/0.952 for double-expressors. Survival analysis demonstrates that for progression-free survival, model-predicted TMA scores significantly stratify double-expressors and non double-expressors (p = 0.0345), whereas pathologist scores do not (p = 0.128). CONCLUSIONS: We conclude that proportion of positive stains can be regressed using attention-based multiple instance learning, that these models generalize well to whole slide images, and that our models can provide non-inferior stratification of progression-free survival outcomes.

Adapting SAM to Histopathology Images for Tumor Bud Segmentation in Colorectal Cancer.

Colorectal cancer (CRC) is the third most common cancer in the United States. Tumor Budding (TB) detection and quantification are crucial yet labor-intensive steps in determining the CRC stage through the analysis of histopathology images. To help with this process, we adapt the Segment Anything Model (SAM) on the CRC histopathology images to segment TBs using SAM-Adapter. In this approach, we automatically take task-specific prompts from CRC images and train the SAM model in a parameter-efficient way. We compare the predictions of our model with the predictions from a trained-from-scratch model using the annotations from a pathologist. As a result, our model achieves an intersection over union (IoU) of 0.65 and an instance-level Dice score of 0.75, which are promising in matching the pathologist's TB annotation. We believe our study offers a novel solution to identify TBs on H&E-stained histopathology images. Our study also demonstrates the value of adapting the foundation model for pathology image segmentation tasks.

Few-shot Tumor Bud Segmentation Using Generative Model in Colorectal Carcinoma.

Current deep learning methods in histopathology are limited by the small amount of available data and time consumption in labeling the data. Colorectal cancer (CRC) tumor budding quantification performed using H&E-stained slides is crucial for cancer staging and prognosis but is subject to labor-intensive annotation and human bias. Thus, acquiring a large-scale, fully annotated dataset for training a tumor budding (TB) segmentation/detection system is difficult. Here, we present a DatasetGAN-based approach that can generate essentially an unlimited number of images with TB masks from a moderate number of unlabeled images and a few annotated images. The images generated by our model closely resemble the real colon tissue on H&E-stained slides. We test the performance of this model by training a downstream segmentation model, UNet++, on the generated images and masks. Our results show that the trained UNet++ model can achieve reasonable TB segmentation performance, especially at the instance level. This study demonstrates the potential of developing an annotation-efficient segmentation model for automatic TB detection and quantification.

Differential diagnosis of frontotemporal dementia subtypes with explainable deep learning on structural MRI.

Background: Frontotemporal dementia (FTD) represents a collection of neurobehavioral and neurocognitive syndromes that are associated with a significant degree of clinical, pathological, and genetic heterogeneity. Such heterogeneity hinders the identification of effective biomarkers, preventing effective targeted recruitment of participants in clinical trials for developing potential interventions and treatments. In the present study, we aim to automatically differentiate patients with three clinical phenotypes of FTD, behavioral-variant FTD (bvFTD), semantic variant PPA (svPPA), and nonfluent variant PPA (nfvPPA), based on their structural MRI by training a deep neural network (DNN). Methods: Data from 277 FTD patients (173 bvFTD, 63 nfvPPA, and 41 svPPA) recruited from two multi-site neuroimaging datasets: the Frontotemporal Lobar Degeneration Neuroimaging Initiative and the ARTFL-LEFFTDS Longitudinal Frontotemporal Lobar Degeneration databases. Raw T1-weighted MRI data were preprocessed and parcellated into patch-based ROIs, with cortical thickness and volume features extracted and harmonized to control the confounding effects of sex, age, total intracranial volume, cohort, and scanner difference. A multi-type parallel feature embedding framework was trained to classify three FTD subtypes with a weighted cross-entropy loss function used to account for unbalanced sample sizes. Feature visualization was achieved through post-hoc analysis using an integrated gradient approach. Results: The proposed differential diagnosis framework achieved a mean balanced accuracy of 0.80 for bvFTD, 0.82 for nfvPPA, 0.89 for svPPA, and an overall balanced accuracy of 0.84. Feature importance maps showed more localized differential patterns among different FTD subtypes compared to groupwise statistical mapping. Conclusion: In this study, we demonstrated the efficiency and effectiveness of using explainable deep-learning-based parallel feature embedding and visualization framework on MRI-derived multi-type structural patterns to differentiate three clinically defined subphenotypes of FTD: bvFTD, nfvPPA, and svPPA, which could help with the identification of at-risk populations for early and precise diagnosis for intervention planning.