Pancreatic ductal adenocarcinoma (PDAC) remains among the most lethal malignancies, with a five-year survival rate below 12%, largely attributable to its asymptomatic onset, late-stage diagnosis, and limited curative treatment options. Although PDAC accounts for approximately 3% of all cancers, it is projected to become the second leading cause of cancer-related mortality in the United States by 2030. A major contributor to its dismal prognosis is the lack of validated early detection strategies for asymptomatic individuals. In this review, we present a comprehensive synthesis of current advances in the early detection of PDAC, with a focus on the identification of high-risk populations, novel biomarker platforms, advanced imaging modalities, and artificial intelligence (AI)-driven tools. We highlight high-risk groups—such as those with new-onset diabetes after age 50, pancreatic steatosis, chronic pancreatitis, cystic precursor lesions, and hereditary cancer syndromes—as priority populations for targeted surveillance. Novel biomarker panels, including circulating tumor DNA (ctDNA), miRNAs, and exosomes, have demonstrated improved diagnostic accuracy in early-stage disease. Recent developments in imaging, such as multiparametric MRI, contrast-enhanced endoscopic ultrasound, and molecular imaging, offer improved sensitivity in detecting small or precursor lesions. AI-enhanced radiomics and machine learning models applied to prediagnostic CT scans and electronic health records are emerging as valuable tools for risk prediction prior to clinical presentation. We further refine the Define–Enrich–Find (DEF) framework to propose a clinically actionable strategy that integrates these innovations. Collectively, these advances pave the way for personalized, multimodal surveillance strategies with the potential to improve outcomes in this historically challenging malignancy.

Objective: To emphasize the importance of pancreatic imaging and the application of artificial intelligence (AI) for enhanced risk prediction of pancreatic ductal adenocarcinoma (PDAC). Background: Detecting PDAC at the early stage is challenging as the disease either remains asymptomatic or presents nonspecific symptoms. Risk prediction of PDAC is an efficient strategy as subsequent targeted screening can assist in diagnosing cancer at the early stage even before the symptoms appear. However, the lack of specific clinical and epidemiological predictors of PDAC makes prediction a highly challenging task. Detecting precursor changes in the pancreas can potentially assist in the risk prediction of PDAC as the precancerous pancreas evolves through biological adaptations-presented as morphological and textural changes on abdominal imaging. However, such microlevel “clues” usually remain unnoticed or unappreciated, partly due to the unavailability of tools to detect and interpret such complex measurements, making the risk prediction of PDAC an unresolved problem. Methods: This review study highlights the limitations of the current risk prediction models of PDAC and the importance of abdominal imaging for predicting PDAC. A suggestive narrative is made as to how recent AI tools can assist in extracting precise measurements of biomarkers, detecting early signs and precancerous abnormalities, quantifying tissue characteristics, and revealing complex features potentially indicative of future incidence of pancreatic cancer (PC) using abdominal imaging. With the help of peer examples of other cancers, a case is built about the application of AI in utilizing image features of the pancreas to enhance risk prediction of PDAC. Furthermore, the challenges of AI applications including insufficient data for model training, risk of data privacy violation, inconsistent data labeling, and limited computational resources, and their potential solutions are also discussed. Conclusions: The recent advancement in the domain of AI is a potential opportunity to utilize automated tools for the identification of imaging-based indicators of PDAC and perform enhanced risk prediction of cancer. With this awareness and motivation, better management of PDAC has expected.

The abdominal computed tomography (CT) scans provide a noninvasive means to assess microlevel morphological features of the pancreas and assists automated diagnosis and treatment plans for several pancreatic disorders. Accurate segmentation of the pancreas in CT images is a prerequisite for reliable quantitative assessment of pancreatic features. Manual outlining of the pancreas can be erroneous, inconsistent, and impractical for large datasets. Automated segmentation is thus highly preferable but is challenged by two major complications: first, the intensity and textural properties of the pancreas and its peripheral organs (e.g., liver, kidney, spleen, etc.) on CT images are highly identical, often resulting in several false positive segmented regions. Second, the morphology of the pancreas is immensely variable and complex, deviating algorithms to specify the exact boundary. This challenge becomes even more severe in the presence of visceral fat tissue around the pancreas that creates fuzzy edges and inconsistent contrast on CT images. All these factors undermine automated segmentation, and therefore, there is a pressing need for a highly robust method that can efficiently overcome these issues currently withholding rigorous assessment of the pancreas.

The accurate segmentation of pancreatic subregions (head, body, and tail) in CT images provides an opportunity to examine the local morphological and textural changes in the pancreas. Quantifying such changes aids in understanding the spatial heterogeneity of the pancreas and assists in the diagnosis and treatment planning of pancreatic cancer. Manual outlining of pancreatic subregions is tedious, time-consuming, and prone to subjective inconsistency. This paper presents a multistage anatomy-guided framework for accurate and automatic 3D segmentation of pancreatic subregions in CT images. Using the delineated pancreas, two soft-label maps were estimated for subregional segmentation—one by training a fully supervised naïve Bayes model that considers the length and volumetric proportions of each subregional structure based on their anatomical arrangement, and the other by using the conventional deep learning U-Net architecture for 3D segmentation. The U-Net model then estimates the joint probability of the two maps and performs optimal segmentation of subregions. Model performance was assessed using three datasets of contrast-enhanced abdominal CT scans: one public NIH dataset of the healthy pancreas, and two datasets D₁ and D₂ (one for each of pre-cancerous and cancerous pancreas). The model demonstrated excellent performance during the multifold cross-validation using the NIH dataset, and external validation using D₁ and D₂. To the best of our knowledge, this is the first automated model for the segmentation of pancreatic subregions in CT images. A dataset consisting of reference anatomical labels for subregions in all images of the NIH dataset is also established.

Accurate staging of bladder cancer assists in identifying optimal treatment (e.g., transurethral resection vs. radical cystectomy vs. bladder preservation). However, currently, about one-third of patients are over-staged and one-third are under-staged. There is a pressing need for a more accurate staging modality to evaluate patients with bladder cancer to assist clinical decision-making. We hypothesize that MRI/RNA-seq-based radiogenomics and artificial intelligence can more accurately stage bladder cancer. A total of 40 magnetic resonance imaging (MRI) and matched formalin-fixed paraffin-embedded (FFPE) tissues were available for analysis. Twenty-eight (28) MRI and their matched FFPE tissues were available for training analysis, and 12 matched MRI and FFPE tissues were used for validation. FFPE samples were subjected to bulk RNA-seq, followed by bioinformatics analysis. In the radiomics, several hundred image-based features from bladder tumors in MRI were extracted and analyzed. Overall, the model obtained mean sensitivity, specificity, and accuracy of 94%, 88%, and 92%, respectively, in differentiating intra- vs. extra-bladder cancer. The proposed model demonstrated improvement in the three matrices by 17%, 33%, and 25% and 17%, 16%, and 17% as compared to the genetic- and radiomic-based models alone, respectively. The radiogenomics of bladder cancer provides insight into discriminative features capable of more accurately staging bladder cancer. Additional studies are underway.

Retinal images contain forests of mutually intersecting and overlapping venous and arterial vascular trees. The geometry of these trees shows adaptation to vascular diseases including diabetes, stroke and hypertension. Segmentation of the retinal vascular network is complicated by inconsistent vessel contrast, fuzzy edges, variable image quality, media opacities, complex intersections and overlaps. This paper presents a Bayesian approach to resolving the configuration of vascular junctions to correctly construct the vascular trees. A probabilistic model of vascular joints (terminals, bridges and bifurcations) and their configuration in junctions is built, and Maximum A Posteriori (MAP) estimation used to select most likely configurations. The model is built using a reference set of 3010 joints extracted from the DRIVE public domain vascular segmentation dataset, and evaluated on 3435 joints from the DRIVE test set, demonstrating an accuracy of 95.2%.