Search

Computer-aided diagnosis (CAD) systems can help radiologists in numerous medical tasks including classification and staging of the various diseases. The 3D tomosynthesis imaging technique adds value to the CAD systems in diagnosis and classification of the breast lesions. Several convolutional neural network (CNN) architectures have been proposed to classify the lesion shapes to the respective classes using a similar imaging method. However, not only is the black box nature of these CNN models questionable in the healthcare domain, but so is the morphological-based cancer classification, concerning the clinicians. As a result, this study proposes both a mathematically and visually explainable deep-learning-driven multiclass shape-based classification framework for the tomosynthesis breast lesion images. In this study, authors exploit eight pretrained CNN architectures for the classification task on the previously extracted regions of interests images containing the lesions. Additionally, the study also unleashes the black box nature of the deep learning models using two well-known perceptive explainable artificial intelligence (XAI) algorithms including Grad-CAM and LIME. Moreover, two mathematical-structure-based interpretability techniques, i.e., t-SNE and UMAP, are employed to investigate the pretrained models’ behavior towards multiclass feature clustering. The experimental results of the classification task validate the applicability of the proposed framework by yielding the mean area under the curve of 98.2%. The explanability study validates the applicability of all employed methods, mainly emphasizing the pros and cons of both Grad-CAM and LIME methods that can provide useful insights towards explainable CAD systems.

On 12 March 2020, the World Health Organization (WHO) declared the novel Coronavirus (CoV) disease a global Pandemic and an emerging risk. In order to understand patterns that are typical in COVID-19 pneumonia and track the evolution of the disease, the role of the chest computed tomography (CT) is pivotal. The impact of the illness as well as the efficiency of the therapy are also monitored carefully when performing this imaging exam. Coronaviruses, specifically CoV-2, as RNA viruses, have a tendency to frequently change their genome, giving the virus beneficial characteristics such as greater transmissibility, pathogenicity and the possibility to escape the previously acquired immunity. Therefore, genome evaluation became an extremely important routine practice worldwide. In particular, in Italy, four variants have been recognised and each of them represent a specific temporal wave of the disease. Hence, our goal was to describe imaging findings of COVID-19 pneumonia, specifically its most typical imaging identified during the period of our study, and to assess whether or not SARS-CoV-2 variants determine different CT patterns. Our analyses revealed that the SARS-CoV-2 genotype seems not to interfere with the severity of CT patterns and, in particular, bilateral Ground Glass Opacities (GGOs) are the most frequent findings in all COVID-19 waves.

Ovarian cancer (OC) is the second most common gynecological malignancy, accounting for about 14,000 deaths in 2020 in the US. The recognition of tools for proper screening, early diagnosis, and prognosis of OC is still lagging. The application of methods such as radiomics to medical images such as ultrasound scan (US), computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET) in OC may help to realize so-called “precision medicine” by developing new quantification metrics linking qualitative and/or quantitative data imaging to achieve clinical diagnostic endpoints. This narrative review aims to summarize the applications of radiomics as a support in the management of a complex pathology such as ovarian cancer. We give an insight into the current evidence on radiomics applied to different imaging methods.

Despite several studies having identified factors associated with successful treatment outcomes in locally advanced cervical cancer, there is the lack of accurate predictive modeling for progression-free survival (PFS) in patients who undergo radical hysterectomy after neoadjuvant chemotherapy (NACT). Here we investigated whether machine learning (ML) may have the potential to provide a tool to predict neoadjuvant treatment response as PFS. In this retrospective observational study, we analyzed patients with locally advanced cervical cancer (FIGO stages IB2, IB3, IIA1, IIA2, IIB, and IIIC1) who were followed in a tertiary center from 2010 to 2018. Demographic and clinical characteristics were collected at either treatment baseline or at 24-month follow-up. Furthermore, we recorded data about magnetic resonance imaging (MRI) examinations and post-surgery histopathology. Proper feature selection was used to determine an attribute core set. Three different machine learning algorithms, namely Logistic Regression (LR), Random Forest (RFF), and K-nearest neighbors (KNN), were then trained and validated with 10-fold cross-validation to predict 24-month PFS. Our analysis included n. 92 patients. The attribute core set used to train machine learning algorithms included the presence/absence of fornix infiltration at pre-treatment MRI as well as of either parametrium invasion and lymph nodes involvement at post-surgery histopathology. RFF showed the best performance (accuracy 82.4%, precision 83.4%, recall 96.2%, area under receiver operating characteristic curve (AUROC) 0.82). We developed an accurate ML model to predict 24-month PFS.

Contrast-Enhanced Spectral Mammography (CESM) is a recently introduced mammographic method with characteristics particularly suitable for breast cancer radiomic analysis. This work aims to evaluate radiomic features for predicting histological outcome and two cancer molecular subtypes, namely Human Epidermal growth factor Receptor 2 (HER2)-positive and triple-negative. From 52 patients, 68 lesions were identified and confirmed on histological examination. Radiomic analysis was performed on regions of interest (ROIs) selected from both low-energy (LE) and ReCombined (RC) CESM images. Fourteen statistical features were extracted from each ROI. Expression of estrogen receptor (ER) was significantly correlated with variation coefficient and variation range calculated on both LE and RC images; progesterone receptor (PR) with skewness index calculated on LE images; and Ki67 with variation coefficient, variation range, entropy and relative smoothness indices calculated on RC images. HER2 was significantly associated with relative smoothness calculated on LE images, and grading tumor with variation coefficient, entropy and relative smoothness calculated on RC images. Encouraging results for differentiation between ER+/ER−, PR+/PR−, HER2+/HER2−, Ki67+/Ki67−, High-Grade/Low-Grade and TN/NTN were obtained. Specifically, the highest performances were obtained for discriminating HER2+/HER2− (90.87%), ER+/ER− (83.79%) and Ki67+/Ki67− (84.80%). Our results suggest an interesting role for radiomics in CESM to predict histological outcomes and particular tumors’ molecular subtype.