Learning from Similarity Proportion Loss for Classifying Skeletal Muscle Recovery Stages

TL;DR

The paper tackles automated classification of skeletal muscle recovery stages from high‑resolution WSIs under weak supervision. It introduces Ordinal Scale Learning from Similarity Proportion (OSLSP), which uses a similarity‑proportion loss derived from two bags to update the backbone while preserving the ordinal relationships among recovery stages. The method defines class similarity with $sim(k,k') = 1 - \frac{|k - k'|}{K-1}$ and constructs ground‑truth PDFs from bag proportions, employing a differentiable Gaussian expansion to enable gradient flow, with a KL‑divergence LLP term for classifier training. On a skeletal muscle recovery dataset, OSLSP achieves higher accuracy than baselines and demonstrates the value of ordinal‑aware, weakly supervised learning for quantitative muscle regeneration analysis.

Abstract

Evaluating the regeneration process of damaged muscle tissue is a fundamental analysis in muscle research to measure experimental effect sizes and uncover mechanisms behind muscle weakness due to aging and disease. The conventional approach to assessing muscle tissue regeneration involves whole-slide imaging and expert visual inspection of the recovery stages based on the morphological information of cells and fibers. There is a need to replace these tasks with automated methods incorporating machine learning techniques to ensure a quantitative and objective analysis. Given the limited availability of fully labeled data, a possible approach is Learning from Label Proportions (LLP), a weakly supervised learning method using class label proportions. However, current LLP methods have two limitations: (1) they cannot adapt the feature extractor for muscle tissues, and (2) they treat the classes representing recovery stages and cell morphological changes as nominal, resulting in the loss of ordinal information. To address these issues, we propose Ordinal Scale Learning from Similarity Proportion (OSLSP), which uses a similarity proportion loss derived from two bag combinations. OSLSP can update the feature extractor by using class proportion attention to the ordinal scale of the class. Our model with OSLSP outperforms large-scale pre-trained and fine-tuning models in classification tasks of skeletal muscle recovery stages.

Figures (2)

• Figure 1: Overview of our OSLSP. (a) Similarity of class and morphological changes of cells over time. (b) A pipeline to obtain instance features and class inference results for each cell from WSI. (c) We computed similarity proportion loss using KL divergence between ground truth similarity distribution derived by combining two bags' proportions and predicted one derived by cosine similarity of each instance in two bags.
• Figure 2: Classification results of WSIs for each day. Blue: ghost fiber, Red: intact myofiber, Pink: recovered myofiber, Orange: myotube, Yellow: myoblast. In the manual expert annotations, unannotated white areas indicate uncertain regions.