The field of medical imaging analysis is moving towards more accurate and efficient methods for image-based profiling and feature extraction. Recent developments focus on improving the accuracy of linear measurements, landmark detection, and shape quantification. A key direction is the integration of deep learning techniques with traditional methods to enhance performance and reduce errors. Noteworthy papers include:
- EnLVAM, which proposes a novel framework for enhancing left ventricle measurement accuracy by enforcing straight-line constraints, and
- TopoNet, which introduces a topology-constrained learning framework for laparoscopic liver landmark detection, and
- ShapeEmbed, which presents a self-supervised learning framework for 2D contour quantification, and
- cp_measure, which provides an API-first feature extraction tool for image-based profiling workflows.