A survey of artificial intelligence/machine learning-based trends for prostate cancer analysis

Abstract

Different types of cancer are more commonly encountered recently. This may be attributed to a variety of reasons, including heredity, changes in the living conditions (food, drinks, pollution, etc.), advancement in technology which allowed for better diagnosis of diseases, among others. Prostate one of the main types of cancers witnessed in males; it has indeed been identified as the second type cancer leading to death in males. Accordingly, it has received considerable attention from the research community where computer scientists and data analysts are closely collaborating with pathologists to develop automated techniques and tools capable of classifying and identifying cancerous cases with high accuracy. These efforts are described in the literature in a large number of research articles which makes it hard and time consuming for researchers to grasp the current state of the art. Instead, review articles form a valuable source for researchers who are interesting in coping with the developments in the field. Generally, the literature includes several survey papers on prostate cancer; each of them tackles some aspect of the domain up to the time when the survey was prepared. Hence the need for the survey described in this paper which highlights the scope of each of the previous survey papers encountered in the literature and adds upon the latest developments in the field as described in more recent papers published mainly in 2023 and 2024. The survey focuses on the main artificial intelligence and machine learning techniques for diagnosing prostate cancer based on various types of data, including MRI. The most recent techniques employed in analyzing prostate cancer data, the various types of data, the available datasets, the reported results, etc. are all covered. This will help researchers in their efforts to keep track of the recent developments in the field and to realize the challenges which need more attention along the path towards developing robust and effect decision support systems for pathologists to have higher self confidence in handling their patients.