Clinical Validation Benchmark Dataset and Expert Performance Baseline for Colorectal Polyp Localization Methods.

Colorectal cancer is one of the leading death causes worldwide, but, fortunately, early detection highly increases survival rates, with the adenoma detection rate being one surrogate marker for colonoscopy quality. Artificial intelligence and deep learning methods have been applied with great success to improve polyp detection and localization and, therefore, the adenoma detection rate. In this regard, a comparison with clinical experts is required to prove the added value of the systems. Nevertheless, there is no standardized comparison in a laboratory setting before their clinical validation. The ClinExpPICCOLO comprises 65 unedited endoscopic images that represent the clinical setting. They include white light imaging and narrow band imaging, with one third of the images containing a lesion but, differently to another public datasets, the lesion does not appear well-centered in the image. Together with the dataset, an expert clinical performance baseline has been established with the performance of 146 gastroenterologists, who were required to locate the lesions in the selected images. Results shows statistically significant differences between experience groups. Expert gastroenterologists' accuracy was 77.74, while sensitivity and specificity were 86.47 and 74.33, respectively. These values can be established as minimum values for a DL method before performing a clinical trial in the hospital setting.

Novel Pixelwise Co-Registered Hematoxylin-Eosin and Multiphoton Microscopy Image Dataset for Human Colon Lesion Diagnosis.

Colorectal cancer presents one of the most elevated incidences of cancer worldwide. Colonoscopy relies on histopathology analysis of hematoxylin-eosin (H&E) images of the removed tissue. Novel techniques such as multi-photon microscopy (MPM) show promising results for performing real-time optical biopsies. However, clinicians are not used to this imaging modality and correlation between MPM and H&E information is not clear. The objective of this paper is to describe and make publicly available an extensive dataset of fully co-registered H&E and MPM images that allows the research community to analyze the relationship between MPM and H&E histopathological images and the effect of the semantic gap that prevents clinicians from correctly diagnosing MPM images. The dataset provides a fully scanned tissue images at 10x optical resolution (0.5 µm/px) from 50 samples of lesions obtained by colonoscopies and colectomies. Diagnostics capabilities of TPF and H&E images were compared. Additionally, TPF tiles were virtually stained into H&E images by means of a deep-learning model. A panel of 5 expert pathologists evaluated the different modalities into three classes (healthy, adenoma/hyperplastic, and adenocarcinoma). Results showed that the performance of the pathologists over MPM images was 65% of the H&E performance while the virtual staining method achieved 90%. MPM imaging can provide appropriate information for diagnosing colorectal cancer without the need for H&E staining. However, the existing semantic gap among modalities needs to be corrected.

Analysis on the Characterization of Multiphoton Microscopy Images for Malignant Neoplastic Colon Lesion Detection under Deep Learning Methods.

BACKGROUND: Colorectal cancer has a high incidence rate worldwide, with over 1.8 million new cases and 880,792 deaths in 2018. Fortunately, its early detection significantly increases the survival rate, reaching a cure rate of 90% when diagnosed at a localized stage. Colonoscopy is the gold standard technique for detection and removal of colorectal lesions with potential to evolve into cancer. When polyps are found in a patient, the current procedure is their complete removal. However, in this process, gastroenterologists cannot assure complete resection and clean margins which are given by the histopathology analysis of the removed tissue, which is performed at laboratory. AIMS: In this paper, we demonstrate the capabilities of multiphoton microscopy (MPM) technology to provide imaging biomarkers that can be extracted by deep learning techniques to identify malignant neoplastic colon lesions and distinguish them from healthy, hyperplastic, or benign neoplastic tissue, without the need for histopathological staining. MATERIALS AND METHODS: To this end, we present a novel MPM public dataset containing 14,712 images obtained from 42 patients and grouped into 2 classes. A convolutional neural network is trained on this dataset and a spatially coherent predictions scheme is applied for performance improvement. RESULTS: We obtained a sensitivity of 0.8228 ± 0.1575 and a specificity of 0.9114 ± 0.0814 on detecting malignant neoplastic lesions. We also validated this approach to estimate the self-confidence of the network on its own predictions, obtaining a mean sensitivity of 0.8697 and a mean specificity of 0.9524 with the 18.67% of the images classified as uncertain. CONCLUSIONS: This work lays the foundations for performing in vivo optical colon biopsies by combining this novel imaging technology together with deep learning algorithms, hence avoiding unnecessary polyp resection and allowing in situ diagnosis assessment.

Deep learning to find colorectal polyps in colonoscopy: A systematic literature review.

Colorectal cancer has a great incidence rate worldwide, but its early detection significantly increases the survival rate. Colonoscopy is the gold standard procedure for diagnosis and removal of colorectal lesions with potential to evolve into cancer and computer-aided detection systems can help gastroenterologists to increase the adenoma detection rate, one of the main indicators for colonoscopy quality and predictor for colorectal cancer prevention. The recent success of deep learning approaches in computer vision has also reached this field and has boosted the number of proposed methods for polyp detection, localization and segmentation. Through a systematic search, 35 works have been retrieved. The current systematic review provides an analysis of these methods, stating advantages and disadvantages for the different categories used; comments seven publicly available datasets of colonoscopy images; analyses the metrics used for reporting and identifies future challenges and recommendations. Convolutional neural networks are the most used architecture together with an important presence of data augmentation strategies, mainly based on image transformations and the use of patches. End-to-end methods are preferred over hybrid methods, with a rising tendency. As for detection and localization tasks, the most used metric for reporting is the recall, while Intersection over Union is highly used in segmentation. One of the major concerns is the difficulty for a fair comparison and reproducibility of methods. Even despite the organization of challenges, there is still a need for a common validation framework based on a large, annotated and publicly available database, which also includes the most convenient metrics to report results. Finally, it is also important to highlight that efforts should be focused in the future on proving the clinical value of the deep learning based methods, by increasing the adenoma detection rate.

Unravelling the effect of data augmentation transformations in polyp segmentation.

PURPOSE: Data augmentation is a common technique to overcome the lack of large annotated databases, a usual situation when applying deep learning to medical imaging problems. Nevertheless, there is no consensus on which transformations to apply for a particular field. This work aims at identifying the effect of different transformations on polyp segmentation using deep learning. METHODS: A set of transformations and ranges have been selected, considering image-based (width and height shift, rotation, shear, zooming, horizontal and vertical flip and elastic deformation), pixel-based (changes in brightness and contrast) and application-based (specular lights and blurry frames) transformations. A model has been trained under the same conditions without data augmentation transformations (baseline) and for each of the transformation and ranges, using CVC-EndoSceneStill and Kvasir-SEG, independently. Statistical analysis is performed to compare the baseline performance against results of each range of each transformation on the same test set for each dataset. RESULTS: This basic method identifies the most adequate transformations for each dataset. For CVC-EndoSceneStill, changes in brightness and contrast significantly improve the model performance. On the contrary, Kvasir-SEG benefits to a greater extent from the image-based transformations, especially rotation and shear. Augmentation with synthetic specular lights also improves the performance. CONCLUSION: Despite being infrequently used, pixel-based transformations show a great potential to improve polyp segmentation in CVC-EndoSceneStill. On the other hand, image-based transformations are more suitable for Kvasir-SEG. Problem-based transformations behave similarly in both datasets. Polyp area, brightness and contrast of the dataset have an influence on these differences.

Constellation Loss: Improving the Efficiency of Deep Metric Learning Loss Functions for the Optimal Embedding of histopathological images.

BACKGROUND: Deep learning diagnostic algorithms are proving comparable results with human experts in a wide variety of tasks, and they still require a huge amount of well-annotated data for training, which is often non affordable. Metric learning techniques have allowed a reduction in the required annotated data allowing few-shot learning over deep learning architectures. AIMS AND OBJECTIVES: In this work, we analyze the state-of-the-art loss functions such as triplet loss, contrastive loss, and multi-class N-pair loss for the visual embedding extraction of hematoxylin and eosin (H&E) microscopy images and we propose a novel constellation loss function that takes advantage of the visual distances of the embeddings of the negative samples and thus, performing a regularization that increases the quality of the extracted embeddings. MATERIALS AND METHODS: To this end, we employed the public H&E imaging dataset from the University Medical Center Mannheim (Germany) that contains tissue samples from low-grade and high-grade primary tumors of digitalized colorectal cancer tissue slides. These samples are divided into eight different textures (1. tumour epithelium, 2. simple stroma, 3. complex stroma, 4. immune cells, 5. debris and mucus, 6. mucosal glands, 7. adipose tissue and 8. background,). The dataset was divided randomly into train and test splits and the training split was used to train a classifier to distinguish among the different textures with just 20 training images. The process was repeated 10 times for each loss function. Performance was compared both for cluster compactness and for classification accuracy on separating the aforementioned textures. RESULTS: Our results show that the proposed loss function outperforms the other methods by obtaining more compact clusters (Davis-Boulding: 1.41 ± 0.08, Silhouette: 0.37 ± 0.02) and better classification capabilities (accuracy: 85.0 ± 0.6) over H and E microscopy images. We demonstrate that the proposed constellation loss can be successfully used in the medical domain in situations of data scarcity.