Finding a Suitable Class Distribution for Building Histological Images Datasets Used in Deep Model Training-The Case of Cancer Detection.

The class distribution of a training dataset is an important factor which influences the performance of a deep learning-based system. Understanding the optimal class distribution is therefore crucial when building a new training set which may be costly to annotate. This is the case for histological images used in cancer diagnosis where image annotation requires domain experts. In this paper, we tackle the problem of finding the optimal class distribution of a training set to be able to train an optimal model that detects cancer in histological images. We formulate several hypotheses which are then tested in scores of experiments with hundreds of trials. The experiments have been designed to account for both segmentation and classification frameworks with various class distributions in the training set, such as natural, balanced, over-represented cancer, and over-represented non-cancer. In the case of cancer detection, the experiments show several important results: (a) the natural class distribution produces more accurate results than the artificially generated balanced distribution; (b) the over-representation of non-cancer/negative classes (healthy tissue and/or background classes) compared to cancer/positive classes reduces the number of samples which are falsely predicted as cancer (false positive); (c) the least expensive to annotate non-ROI (non-region-of-interest) data can be useful in compensating for the performance loss in the system due to a shortage of expensive to annotate ROI data; (d) the multi-label examples are more useful than the single-label ones to train a segmentation model; and (e) when the classification model is tuned with a balanced validation set, it is less affected than the segmentation model by the class distribution of the training set.

Evaluation of eight melanocytic and neural crest-associated markers in a well-characterised series of 124 malignant peripheral nerve sheath tumours (MPNST): useful to distinguish MPNST from melanoma?

AIMS: The diagnosis of malignant peripheral nerve sheath tumour (MPNST) may be challenging, especially in the sporadic setting. Owing to the lack of specific histological criteria, immunohistochemical and molecular diagnostic markers, several differential diagnoses must be considered, especially melanoma. Indeed, although S100 protein usually stains melanoma, other melanocytic markers are often negative, especially in spindle cell/desmoplastic types. This pattern of immunoreactivity resembles that of some nerve-derived tumours such as MPNST. Owing to their different clinical behaviours and therapeutic implications, accurate identification of these two different tumours is crucial. METHODS AND RESULTS: S100, SOX10, KBA62, MITF, HMB45, Melan-A, tyrosinase PNL2 and BRAF-V600E immunostaining was performed in a pathologically and genetically well-characterised cohort of primary MPNST (n = 124), including 66 (53%) NF1-associated tumours. Sox10 and KBA62 expression were found, respectively, in 102 (84%) and in 101 (83%) MPNST, whereas S100 was expressed in 64 cases (52%). We observed an increased loss of S100 with increasing histological grade (P = 0.0052). We found Melan-A expression in 14% (n = 17) of all MPNST, occurring in 82% (n = 14) of cases in an NF1 context. Six per cent (n = 8) of MPNST showed tyrosinase positivity, including seven (87%) NF1-associated. MITF expression was found in 10 (8%) MPNST. None expressed PNL2, HMB45 or BRAF-V600E. CONCLUSION: MPNST (in NF1 and a sporadic setting) can quite often be positive for Melan-A, tyrosinase and MITF. Pathologists should be cognisant of these exceptions to prevent confusion with melanoma.