arXiv subjects

Nitin Singhal

Publications and source records attributed to Nitin Singhal.

7 recordsLinked to original sources

arXiv2026

Mitosis Detection in the Wild: Multi-Tumor and Context-Aware Generalization in the MIDOG 2025 Challenge

Automated mitosis detection is a well-established task in computational pathology. While previous benchmarks focused on scanner-induced domain shift, clinical "real-world" application requires models to be robust across the vast variance to be expected in the histological landscape. The MItosis DOmain Generalization (MIDOG) 2025 challenge was designed to evaluate algorithmic performance across unprecedented biological and contextual diversity. We curated a test dataset of 365 cases, encompassing 12 distinct human, canine and feline tumor types, digitized across multiple scanning platforms. Moving beyond hand-selected hotspots, the challenge required detection also in random tissue areas (representative of the whole slide detection situation) and challenging areas (areas rich in hard negatives). In the second track, we introduced the classification of atypical mitotic figures (AMFs). There were 18 teams submitting to the detection track, with F1 scores ranging up to 0.740. In the AMF detection track, we had 21 submissions with balanced accuracy values up to 0.908. Our analysis reveals that while most models perform reliably in traditional hotspots, significant performance degradation occurs in challenging ROIs, where false positive rates tripled. Furthermore, performance varied significantly across the 12 tumor types, highlighting "blind spots" in current state-of-the-art architectures when encountering rare or highly pleomorphic malignancies. Moreover, we evaluated the effectiveness of ensembling and found a mean increases of 1.5 and 1.3 percentage points in F1 score and balanced accuracy, respectively. In contrast, TTA showed no relevant improvement. MIDOG 2025 demonstrates that "in the wild" mitosis detection remains a significant hurdle. The transition from hotspot-only evaluation to a multi-contextual framework provides a more realistic proxy for clinical reliability.

cs.CV↗

arXiv2025

Ensemble YOLO Framework for Multi-Domain Mitotic Figure Detection in Histopathology Images

The reliable identification of mitotic figures in whole-slide histopathological images remains difficult, owing to their low prevalence, substantial morphological heterogeneity, and the inconsistencies introduced by tissue processing and staining procedures. The MIDOG competition series provides standardized benchmarks for evaluating detection approaches across diverse domains, thus motivating the development of generalizable deep learning models. In this work, we investigate the performance of two modern one-stage detectors, YOLOv5 and YOLOv8, trained on MIDOG++, CMC, and CCMCT datasets. To enhance robustness, training incorporated stain-invariant color perturbations and texture-preserving augmentations. Ininternal validation, YOLOv5 achieved higher precision (84.3%), while YOLOv8 offered improved recall (82.6%), reflecting architectural trade-offs between anchor-based and anchor-free detections. To capitalize on their complementary strengths, weemployed an ensemble of the two models, which improved sensitivity (85.3%) while maintaining competitive precision, yielding the best F1 score of 83.1%. On the preliminary MIDOG 2025 test leaderboard, our ensemble ranked 5th with an F1 score of 79.2%, precision of 73.6%, and recall of 85.8%, confirming that the proposed strategy generalizes effectively across unseen test data. These findings highlight the effectiveness of combining anchor-based and anchor-free object detectors to advance automated mitosis detection in digital pathology.

eess.IV↗

arXiv2024

AutoPET III Challenge: Tumor Lesion Segmentation using ResEnc-Model Ensemble

Positron Emission Tomography (PET) /Computed Tomography (CT) is crucial for diagnosing, managing, and planning treatment for various cancers. Developing reliable deep learning models for the segmentation of tumor lesions in PET/CT scans in a multi-tracer multicenter environment, is a critical area of research. Different tracers, such as Fluorodeoxyglucose (FDG) and Prostate-Specific Membrane Antigen (PSMA), have distinct physiological uptake patterns and data from different centers often vary in terms of acquisition protocols, scanner types, and patient populations. Because of this variability, it becomes more difficult to design reliable segmentation algorithms and generalization techniques due to variations in image quality and lesion detectability. To address this challenge, We trained a 3D Residual encoder U-Net within the no new U-Net framework, aiming to generalize the performance of automatic lesion segmentation of whole body PET/CT scans, across different tracers and clinical sites. Further, We explored several preprocessing techniques and ultimately settled on using the Total Segmentator to crop our training data. Additionally, we applied resampling during this process. During inference, we leveraged test-time augmentations and other post-processing techniques to enhance tumor lesion segmentation. Our team currently hold the top position in the Auto-PET III challenge and outperformed the challenge baseline model in the preliminary test set with Dice score of 0.9627.

eess.IV↗

arXiv2022

FUSION: Fully Unsupervised Test-Time Stain Adaptation via Fused Normalization Statistics

Staining reveals the micro structure of the aspirate while creating histopathology slides. Stain variation, defined as a chromatic difference between the source and the target, is caused by varying characteristics during staining, resulting in a distribution shift and poor performance on the target. The goal of stain normalization is to match the target's chromatic distribution to that of the source. However, stain normalisation causes the underlying morphology to distort, resulting in an incorrect diagnosis. We propose FUSION, a new method for promoting stain-adaption by adjusting the model to the target in an unsupervised test-time scenario, eliminating the necessity for significant labelling at the target end. FUSION works by altering the target's batch normalization statistics and fusing them with source statistics using a weighting factor. The algorithm reduces to one of two extremes based on the weighting factor. Despite the lack of training or supervision, FUSION surpasses existing equivalent algorithms for classification and dense predictions (segmentation), as demonstrated by comprehensive experiments on two public datasets.

eess.IV↗

arXiv2022

ADAM Challenge: Detecting Age-related Macular Degeneration from Fundus Images

Age-related macular degeneration (AMD) is the leading cause of visual impairment among elderly in the world. Early detection of AMD is of great importance, as the vision loss caused by this disease is irreversible and permanent. Color fundus photography is the most cost-effective imaging modality to screen for retinal disorders. Cutting edge deep learning based algorithms have been recently developed for automatically detecting AMD from fundus images. However, there are still lack of a comprehensive annotated dataset and standard evaluation benchmarks. To deal with this issue, we set up the Automatic Detection challenge on Age-related Macular degeneration (ADAM), which was held as a satellite event of the ISBI 2020 conference. The ADAM challenge consisted of four tasks which cover the main aspects of detecting and characterizing AMD from fundus images, including detection of AMD, detection and segmentation of optic disc, localization of fovea, and detection and segmentation of lesions. As part of the challenge, we have released a comprehensive dataset of 1200 fundus images with AMD diagnostic labels, pixel-wise segmentation masks for both optic disc and AMD-related lesions (drusen, exudates, hemorrhages and scars, among others), as well as the coordinates corresponding to the location of the macular fovea. A uniform evaluation framework has been built to make a fair comparison of different models using this dataset. During the challenge, 610 results were submitted for online evaluation, with 11 teams finally participating in the onsite challenge. This paper introduces the challenge, the dataset and the evaluation methods, as well as summarizes the participating methods and analyzes their results for each task. In particular, we observed that the ensembling strategy and the incorporation of clinical domain knowledge were the key to improve the performance of the deep learning models.

eess.IV↗

arXiv2020

Histopathological Stain Transfer using Style Transfer Network with Adversarial Loss

Deep learning models that are trained on histopathological images obtained from a single lab and/or scanner give poor inference performance on images obtained from another scanner/lab with a different staining protocol. In recent years, there has been a good amount of research done for image stain normalization to address this issue. In this work, we present a novel approach for the stain normalization problem using fast neural style transfer coupled with adversarial loss. We also propose a novel stain transfer generator network based on High-Resolution Network (HRNet) which requires less training time and gives good generalization with few paired training images of reference stain and test stain. This approach has been tested on Whole Slide Images (WSIs) obtained from 8 different labs, where images from one lab were treated as a reference stain. A deep learning model was trained on this stain and the rest of the images were transferred to it using the corresponding stain transfer generator network. Experimentation suggests that this approach is able to successfully perform stain normalization with good visual quality and provides better inference performance compared to not applying stain normalization.

eess.IV↗

arXiv2017

A Two-Parameter Model for Ultrasonic Tissue Characterization with Harmonic Imaging

Over the past few decades, researchers have developed several approaches such as the Reference Phantom Method (RPM) to estimate ultrasound attenuation coefficient (AC) and backscatter coefficient (BSC). AC and BSC can help to discriminate pathology from normal tissue during in-vivo imaging. In this paper, we propose a new RPM model to simultaneously compute AC and BSC for harmonic imaging and a normalized score that combines the two parameters as a measure of disease progression. The model utilizes the spectral difference between two regions of interest, the first, a proximal, close to the probe and second, a distal, away from the probe. We have implemented an algorithm based on the model and shown that it provides accurate and stable estimates to within 5% of AC and BSC for simulated received echo from post-focal depths of a homogeneous liver-like medium. For practical applications with time gain and time frequency compensated in-phase and quadrature (IQ) data from ultrasound scanner, the method has been approximated and generalized to estimate AC and BSC for tissue layer underlying a more attenuative subcutaneous layer. The angular spectrum approach for ultrasound propagation in biological tissue is employed as a virtual Reference Phantom (VRP). The VRP is calibrated with a fixed probe and scanning protocol for application to liver tissue. In a feasibility study with 16 subjects, the method is able to separate 9/11 cases of progressive non-alcoholic fatty liver disease from 5 normal. In particular, it is able to separate 4/5 cases of non-alcoholic steato-hepatitis and early fibrosis (F<=2) from normal tissue. More extensive clinical studies are needed to assess the full capability of this model for screening and monitoring disease progression in liver and other tissues.

physics.med-ph↗