A Machine Learning and Radiomics Approach in Lung Cancer for Predicting Histological Subtype

Abstract

Lung cancer is one of the deadliest diseases worldwide. Computed Tomography (CT) images are a powerful tool for investigating the structure and texture of lung nodules. For a long time, trained radiologists have performed the grading and staging of cancer severity by relying on radiographic images. Recently, radiomics has been changing the traditional workflow for lung cancer staging by providing the technical and methodological means to analytically quantify lesions so that more accurate predictions could be performed while reducing the time required from each specialist to perform such tasks. In this work, we implemented a pipeline for identifying a radiomic signature composed of a reduced number of features to discriminate between adenocarcinomas and other cancer types. In addition, we also investigated the reproducibility of this radiomic study analysing the performances of the classification models on external validation data. In detail, we first considered two publicly available datasets, namely D1 and D2, composed of n = 262 and n = 89 samples, respectively. Ten significant features, according to univariate AUC evaluated on D1, were retained. Mann–Whitney U tests recognised three of these features to have a statistically different distribution, with a p-value < 0.05. Then, we collected n = 51 CT images from patients with lung nodules at the Azienda Ospedaliero—Universitaria “Policlinico Riuniti” in Foggia. Resident radiologists manually annotated the lung lesions in images to allow the subsequent analysis of the malignancy regions. We designed a pipeline for feature extraction from the Volumes of Interest in order to generate a third dataset, i.e., D3. Several experiments have been performed showing that the selected radiomic signature not only allowed the discrimination of lung adenocarcinoma from other cancer types independently from the input dataset used for training the models, but also allowed reaching good classification performances also on external validation data; in fact, the radiomic signature computed on D1 and evaluated on the local cohort allowed reaching an AUC of 0.70 (p<0.001) for the task of predicting the histological subtype.

Tumor Cellularity Assessment of Breast Histopathological Slides via Instance Segmentation and Pathomic Features Explainability

“…In the first approach, we performed the tumor/non-tumor classification of the cell nuclei taking advantage of the features extracted by the DL detection models. In the second case, quantitative descriptors of cell shape, morphology, and texture, i.e., pathomic features, are extracted using the PyRadiomics library [ 15 ], which offers a systematic pipeline for easing the quantitative analysis of medical images, with several applications in the radiomics field [ 16 , 17 , 18 ], but more recently for pathomics as well [ 19 ]. Both workflows are schematized in Figure 2 .…”

A Fusion Biopsy Framework for Prostate Cancer Based on Deformable Superellipses and nnU-Net

“…The semantic segmentation of the prostate gland from MRI can be efficiently met via DL techniques, as fully convolutional neural networks [ 24 ]. Semantic segmentation, which poses the basis for subsequent classification and characterization tasks [ 25 , 26 ], is essential in numerous clinical applications including artificial intelligence in diagnostic support systems, therapy planning, intraoperative assistance, and monitoring of tumor growth. It is a computer vision task that can be computed with DL algorithms and consists of labeling each pixel of an input image, without recognizing the different instances of objects [ 27 , 28 ]; it is possible to see semantic segmentation as a problem of conversion from image to image, where the input image is the original image and each pixel intensity value of the output image indicates the relation of that pixel to the associated class [ 29 ].…”

Segmentation of Variants of Nuclei on Whole Slide Images by Using Radiomic Features

“…To address this problem, we employed radiomics (quantitative) features that help to extract high-throughput information [ 2 , 3 ] from the patches of WSIs [ 4 ], and subsequent modeling using machine, and deep learning algorithms [ 5 ]. Existing learning methods exhibit considerable potential for solving general nuclei segmentation, but obtaining distinct and inferior quality multiple variants of nuclei from WSIs of different organs of the body using deep learning models is difficult because customized parameters are required for each experiment, preparing respective annotations is time-consuming, and/or configuration of training algorithms is difficult [ 6 , 7 , 8 ].…”