Abstract PO5-14-04: FTIR Spectroscopy Analysis of Lipid Region: Distinguishing Breast Cancer from Benign Breast Diseases Using PCA-SVM

Background: Breast cancer (BC) is the most diagnosed type worldwide, with 2.26 million new cases in 2020. Mammography is the preferred screening method, but it has limitations such as radiation exposure and low sensitivity in dense breasts. Therefore, there is a need for accurate diagnostic methods to detect breast cancer and distinguish benign conditions. Objective: This study aimed to develop an automated tool for the comparison of lipid spectral region in benign breast disease (BBD) and breast cancer (BC) to determine their potential diagnostic value using Fourier Transform Infrared Spectroscopy (ATR-FTIR). Methods: This study was conducted at a Clinical Hospital in Uberlandia, MG, Brazil, after approval by the Human Research Ethics Committee of the Federal University of Uberlandia, and all subjects provided written informed consent. Women who went to the Clinical Hospital for breast surgery were invited to participate in the study as volunteers. After surgery and histopathological analysis, the tumors and lesions found were classified according to histological type, staging, and the status of ER, PR, and HER2 receptors. A total of 60 women participated in the study, of whom 27 had BBD, and 33 had BC. Spectra were measured in the wavenumber range of 4000 cm-1 to 650 cm-1 using an Agilent Cary 600 Series FTIR spectrometer coupled to an MCT detector. The air spectrum was used as a background before each sample analysis. The sample spectra were obtained in triplicate, with a spectral resolution of 4 cm-1, and 128 scans were performed for each measurement. The infrared spectra were analyzed after undergoing preprocessing, which involved positive baseline rubberband normalization, normalization by minimum and maximum, and application of the second derivative. Results: The region of 3050-2800 cm-1 was able to differentiate BC from BBD using principal component analysis (PCA) followed by the support vector machine algorithm (PCA-SVM) with an accuracy of 80%, sensitivity of 88%, and specificity of 70%. This range corresponds to the vibration of lipids, and it is known that changes in the contents of biomolecules in the serum, an increase or decrease, can be related to the presence or absence of BC. The implementation of this approach in clinical practice may automate the diagnosis of BC, monitoring whether changes in therapy or interventions are necessary throughout the treatment. FTIR can potentially assist in clinical decision-making regarding follow-up or biopsy recommendations, avoiding unnecessary biopsies that are more invasive and expensive than blood collection, and which often cause significant anxiety to the patient even if the lesion is not highly suspicious. Conclusion: This study shows that it is possible to detect BC with good accuracy through a minimally invasive, fast, and cost-effective method analyzing the lipid spectral region using principal component analysis (PCA) followed by the support vector machine algorithm (PCA-SVM). Citation Format: Juliana Pereira, Alinne Faria, Izabella Ferreira, Letícia Santos, Donizeti Santos, Marcelo Maia, Ohanna Costa, Raul Freitas, Carlos Paiva, Yara Maia. FTIR Spectroscopy Analysis of Lipid Region: Distinguishing Breast Cancer from Benign Breast Diseases Using PCA-SVM [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-14-04.