There are different types of GB disease. From gallstones to cancer, they all have similar symptoms, but they vary widely in severity [1]. All scientists confirm that although the exact causes of GB cancer are unknown, certain factors may increase a person’s risk of developing GB cancer. Frequently, these factors are related to the simple inflammation of the GB. According to the American Cancer Society (ACS) [2] and the American Society of Clinical Oncology (ASCO) [3], the risk of GB cancer is around five times higher in people who have a history of GB conditions, mainly gallstones, compared to those who do not. To this end, it is crucial to diagnose the type of GB disease and how serious it is at an early stage in order to prevent or reduce the wider spread of the disease.

Early detection and diagnosis are the primary sources of lifesaving and are among the most-challenging features of health surveillance. Indeed, according to the Canadian Cancer Society (CCS), about 19% of people diagnosed with GB cancer survive for at least 5 years. Unfortunately, only about 4% of those with stage 4 survive their cancer for 5 or more years, compared to 50% of those diagnosed with stage 1.

To perfectly diagnose GB diseases, ultrasound imaging, as one of the most frequently used imaging modalities, is recognized as a powerful and universal screening and diagnostic tool for physicians and radiologists [4]. After screening, an accurate diagnosis is necessary to identify an appropriate disease treatment plan. Typically, diagnostic information is collected from the patient’s history and clinical examination. Many indications and symptoms can be ambiguous, especially those related to GB diseases, which have similar symptoms. Therefore, ultrasound images (UI) have to be interpreted and understood by highly qualified medical professionals. Given that diagnosis via ultrasound imaging can be time- and labor-consuming, it can be difficult to fund and take advantage of this service in remote locations [5]. Regretfully, many analysts refuse to work in rural areas, and some hospitals do not have the resources to train existing medical professionals to provide this service. As a result, an informative method is needed to simplify the UI acquisition and evaluation process in order to recognize the organ-related pathologies and anomalies in a widely accessible and achievable manner. In rural locations, this informative method can be very useful in computerized healthcare structures, which is the prime motivation of the proposed study. Indeed, if any anomalies are discovered during the initial screening step, specialists and radiologists can detect intraabdominal organ issues and provide the exact treatment. Unskilled radiologists can also use the produced healthcare model to build relative research on scans for evaluating different solutions [6].

Currently, artificial intelligence (AI) techniques, ranging from machine learning (ML) to deep learning (DL), are prevalent in healthcare for disease diagnosis, drug discovery, drug development, and patient risk identification [7]. The advances in DL and deep neural network (DNN)-based methods of research and development provide significant progress in the domain of medical image analysis and understanding [8]. Moreover, with the progress at the algorithmic level as well as the availability of high-performance computing machines and large quantities of data, DL-based methods have become increasingly popular [9]. They are now considered to be the most commonly used and most sophisticated algorithms for handling many computer vision tasks [10]. In addition, DL algorithms are capable of assisting analysts in the early identification, treatment, and recognition of diseases, and they, subsequently, provide efficient methods for medical diagnostics. Indeed, DL algorithms can directly process and automatically learn mid-level and high-level abstract features acquired from immense quantities of raw collected data, in which higher-level abstract features are defined by combining them with lower-level features, to achieve an acceptable level of accuracy and, eventually, to perform automatic UI analysis tasks, such as classification, organ segmentation, and object detection [9,11,12,13,14].