Task 1: Cell level staining pattern classification

For the cell level staining pattern classification (Task 1), we utilized texture features such as rotational invariant co-occurrence (RIC) versions of the well-known local binary pattern (LBP), median binary pattern (MBP), joint adaptive median binary pattern (JAMBP), and motif labels (ML) along with other optimized features. We report the classification results utilizing different classifiers such as the k-nearest neighbors (kNN), support vector machine (SVM), and random forest (RF). We obtained the best accuracy of 94.26% for six cell classes with RIC-LBP combined with a motif co-occurrence labels (MCL).

Task 2: Specimen level staining pattern classification

For specimen level staining pattern classification (Task 2) we utilize a combination RIC-LBP with RF classifier and obtained 80% accuracy for seven classes.

Task 4: Cell segmentation

For cell segmentation (Task 4), we use our optimized multiscale spatial feature bank along with RF classifier for pixelwise labeling and obtained F-measure of 84.26% for 1008 images.

Example segmentations from our pipeline

(Top to bottom: Input, ground-truth (GT), our results)