Unsupervised Learning Based on Multiple Descriptors for WSIs Diagnosis

Abstract

An automatic pathological diagnosis is a challenging task because histopathological images with different cellular heterogeneity representations are sometimes limited. To overcome this, we investigated how the holistic and local appearance features with limited information can be fused to enhance the analysis performance. We propose an unsupervised deep learning model for whole-slide image diagnosis, which uses stacked autoencoders simultaneously feeding multiple-image descriptors such as the histogram of oriented gradients and local binary patterns along with the original image to fuse the heterogeneous features. The pre-trained latent vectors are extracted from each autoencoder, and these fused feature representations are utilized for classification. We observed that training with additional descriptors helps the model to overcome the limitations of multiple variants and the intricate cellular structure of histopathology data by various experiments. Our model outperforms existing state-of-the-art approaches by achieving the highest accuracies of 87.2 for ICIAR2018, 94.6 for Dartmouth, and other significant metrics for public benchmark datasets. Our model does not rely on a specific set of pre-trained features based on classifiers to achieve high performance. Unsupervised spaces are learned from the number of independent multiple descriptors and can be used with different variants of classifiers to classify cancer diseases from whole-slide images. Furthermore, we found that the proposed model classifies the types of breast and lung cancer similar to the viewpoint of pathologists by visualization. We also designed our whole-slide image processing toolbox to extract and process the patches from whole-slide images.

The future of artificial intelligence in thoracic surgery for non-small cell lung cancer treatment a narrative review

“…In the aftermath of lung cancer surgery, the postoperative period unfolds as a critical phase where AI demonstrates its potential in pathological assessment. A multitude of studies, such as those by Sheikh ( 36 ) and DiPalma et al. ( 37 ), accentuate the role of DL methodologies in detailed histological subtyping.…”

“…Sheikh et al. ( 36 ) explored the impact of multiple descriptors on a deep learning model’s performance in the multi-class classification of WSIs. They found that augmenting inputs enhanced the discriminatory capabilities of the model.…”

“…A salient concern extracted from the systematic review is the conspicuous lack of interpretability and explanation within certain AI applications ( 47 ). While the proficiency of AI models in classification tasks is evident, the paucity of concerted efforts in addressing common sense reasoning, especially in deciphering the intricate physical characteristics of cells, poses a critical challenge ( 36 ). This interpretability gap, particularly pronounced in tasks requiring nuanced understanding, introduces a layer of complexity in the integration of AI insights into the clinical decision-making process.…”

“…These studies looked into the complexities of a 5-class problem, addressing various histological patterns encompassing lepidic, acinar, papillary, micropapillary, and solid patterns. Sheikh et al (36) explored the impact of multiple descriptors on a deep learning model's performance in the multi-class classification of WSIs. They found that augmenting inputs enhanced the discriminatory capabilities of the model.…”

“…The depth of learning algorithms, highlighted by multiple studies (8,9,36,39,(41)(42)(43) necessitates substantial volumes of labeled data for effective performance. However, the pragmatic challenge arises when dealing with the sheer scale of annotations required, often dependent on the expertise of pathologists (37).…”