Medical Image Analysis

Medical Image Analysis

Since the discovery of the X-ray radiation by Wilhelm Conrad Roentgen in 1895, the field of medical imaging has developed into a huge scientific discipline. The analysis of patient data acquired by current image modalities, such as computerized tomography (CT), magnetic resonance tomography (MRT), positron emission tomography (PET), or ultrasound (US), offers previously unattained opportunities for diagnosis, therapy planning, and therapy assessment. Medical image processing is essential to leverage this increasing amount of data and to explore and present the contained information in a way suitable for the specific medical task. In this tutorial, we will approach the analysis and visualization of medical image data in an explorative manner. In particular, we will visually construct the image processing algorithms using the popular graphical data-flow builder MeVisLab, which is available as a free download for noncommercial research. We felt that it could be more interesting for the reader to see and explore examples of medical image processing that go beyond simple image enhancements. The part of exploration, to inspect medical image data and experiment with image-processing pipelines, requires software that encourages this kind of visual exploration.

Weakly-supervised Kidney Tumor Classification from CT Scans with Multi-Instance Learning and Anatomical Filtering

Deep learning models for CT scan analysis are often limited by the scarcity of precise pixel-level annotations, which require significant radiologist effort to produce. Training on scan-level labels alone reduces annotation requirements but introduces challenges: low supervision ratios and large input volumes make models prone to overfitting and shortcut learning. In this work, we investigate two complementary methods to address these challenges: multi-instance learning (MIL) and anatomical filtering. MIL divides CT volumes into 2D slice instances, enabling efficient 2D architectures with ImageNet pretraining rather than computationally demanding 3D models. Anatomical filtering uses Compass, our self-supervised body part regression model, to crop scans to pathology-relevant subregions without requiring segmentation masks. We evaluate two MIL frameworks - Attention-based MIL (ABMIL) and FocusMIL - on kidney tumor classification across one internal dataset (TUH) and two external datasets (KiTS23 and TCGA-KiRC). Our best models achieve F1 = 0.83 on the internal test set using only scan-level labels. We further show that anatomical filtering with the Compass model is critical for the out-of-distribution generalization of embedding-based ABMIL, while instance-based FocusMIL demonstrates greater inherent robustness to distribution shift. While evaluated on kidney tumors, we consider this a proof-of-concept for a broader weakly supervised CT classification pipeline applicable to other organs and pathologies.

Medical image segmentation with optimal learning from limited data and annotations: A comprehensive review

This survey extends beyond traditional and deep learning-based augmentation techniques or deep semi-supervised approaches, by explicitly focusing on medical/clinical imaging modalities, by explicitly focusing on CT, MRI, and X-ray, offering a broader perspective.

Uncertainty Quantification in Medical Image Segmentation: A Comprehensive Survey

A comprehensive survey of UQ techniques in medical image segmentation is presented, categorizing existing approaches into Bayesian methods, deep ensembles, deterministic methods, test-time data augmentation, and hybrid models, while treating foundation-model-based UQ as a separate cross-cutting category.

On Cross-Validation for Hyperparameter Optimization of Deep Learning Image Classifiers

Hyperparameter optimization (HPO) can materially affect the performance of deep learning (DL) image classifiers, but there is little empirical guidance on how to derive the validation signal that drives it, especially for the small sample sizes common in fields such as medical imaging. We compared three HPO protocols in terms of {\em absolute performance-estimation error} (AEE; the absolute difference between the winning configuration's validation AUROC and its test AUROC): fixed holdout (F), reshuffled holdout (R), and 5-fold cross-validation (C). The search space, sampler, training procedure, architecture, and test set were held identical across protocols. We evaluated the protocols on three public datasets spanning two regimes: binary medical imaging (RSNA pneumonia radiographs and binarized HAM10000 skin lesions) and 200-class natural imaging (Tiny ImageNet), across a range of development set sizes $n$ and two backbones (ResNet-18 on all datasets, Vision Transformer (ViT-S/16) on RSNA). On the medical datasets, every point estimate favored cross-validation over both holdout protocols, with reductions in AEE largest at small sample sizes and diminishing as $n$ increased. This pattern remained robust under conservative family-wise adjustment. On Tiny ImageNet, AEE was negligible under all three protocols. Test AUROC was generally similar among protocols. Fixed holdout had lower mean AEE than reshuffled holdout in 11 of 12 medical conditions, although this secondary finding was less uniformly supported. For small-sample medical image classification, we recommend cross-validation-based HPO when computational resources permit because it trades additional computation for a more reliable development-time estimate of subsequent test performance.

INTELIGENCIA ARTIFICIAL