Article Highlights
- A recent study led by Tomasz P. Wrobel from Jagiellonian University introduces a classification model utilizing IR spectroscopy to aid in detecting pancreatic cancer.
- The study focuses on pancreatic intraepithelial neoplasia (PanIN), a critical stage in pancreatic cancer progression, highlighting limitations in current detection methods and the importance of accurate grading and characterization for improved patient outcomes.
- Utilizing over 600 biopsies from 250 patients, researchers developed a PLSR model and employed a Random Forest model to detect PanINs, achieving accurate classification.
The pancreas is mainly responsible for two things: to make juices that help digest food, and to make hormones that can help control blood sugar levels (1). Ensuring the health of the pancreas is critical. Pancreatic cancer is a disease where malignant cells form in the tissues of the pancreas (1). According to the American Cancer Society, approximately 66,440 people are estimated to be diagnosed with pancreatic cancer based on past trends, and over 51,000 of those will die from the disease (2).
A recent study from Jagiellonian University, led by Tomasz P. Wrobel, explored using a newly developed classification model to assist clinicians in detecting pancreatic cancer. The results of the study were published in Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy (3).
Pancreatic intraepithelial neoplasia (PanIN) represents a critical stage in pancreatic cancer progression, with high-grade PanINs significantly increasing the risk of developing invasive pancreatic ductal adenocarcinoma (PDAC) (3). However, the current methods for PanIN detection and classification have limitations (3). As a result, it is essential that new approaches are developed to improve the detection of PanIN.
In their study, the researchers utilized infrared (IR) spectroscopy to construct a classification model for PanIN detection (3). Leveraging data from over 600 biopsies collected from 250 patients, they developed a partial least squares regression (PLSR) model capable of characterizing pancreatic ducts from benign to cancerous states (3).
The study was designed to try and understand more fully the pathology progression of pancreatic tissue. To improve patient outcomes, PanIN needs to be accurately graded and characterized (3).
The research team used a Random Forest model to detect PanINs, which they were able to demonstrate as an effective model. The study results show that Random Forest achieved accuracy in distinguishing between different malignancy levels (3). Moreover, their analysis revealed significant biochemical similarities between PanIN and other pathological classes, providing deeper insights into the disease's molecular characteristics (3).
By refining the classification and characterization of PanINs, clinicians can better identify patients at heightened risk of developing aggressive pancreatic cancer, enabling timely interventions and personalized therapeutic strategies (3). As efforts to combat pancreatic cancer intensify, using spectroscopic techniques, such as IR spectroscopy, was demonstrated to have potential in improving diagnostic and therapeutic strategies in oncology.
This article was written with the help of artificial intelligence and has been edited to ensure accuracy and clarity. You can read more about our policy for using AI here.
References
(1) National Cancer Institute, Pancreatic Cancer Treatment (PDQ®)–Patient Version. Available at: https://www.cancer.gov/types/pancreatic/patient/pancreatic-treatment-pdq#:~:text=Key%20Points,is%20difficult%20to%20diagnose%20early. (accessed 2024-03-01).
(2) American Cancer Society, Key Statistics for Pancreatic Cancer. Available at: https://www.cancer.org/cancer/types/pancreatic-cancer/about/key-statistics.html#:~:text=About%2066%2C440%20people%20(34%2C530%20men,will%20die%20of%20pancreatic%20cancer. (accessed 2024-03-01).
(3) Liberda-Matyja, D.; Koziol-Bohatkiewicz, P.; Wrobel, T. P. Pancreatic Intraepithelial Neoplasia Detection and Duct Pathology Grading Using FT-IR Imaging and Machine Learning. Spectrochimica Acta Part A: Mol. Biomol. Spectrosc. 2024, 309, 123756. DOI: 10.1016/j.saa.2023.123756
