Differential Diagnosis of Breast Cancer and Ovarian Cancer Based on ATR-FTIR Spectroscopy Coupled With Machine Learning
SONG Ao1, 2, CAI Yi-sa1, 2, CAI Li-zheng1, 4, YANG Wan-li1, PANG Nan1, YU Rui-hua1, WANG Shi-yan3*, YANG Chao1*, JIANG Feng1*
1. Chongming Hospital Affiliated to Shanghai University of Health and Medicine Sciences, Shanghai 202150, China
2. School of Materials and Chemical Engineering, Huaiyin Institute of Technology, Huaian 223003, China
3. Faculty of Life Science and Food Engineering, Huaiyin Institute of Technology, Huaian 223003, China
4. Shanghai University of Traditional Chinese Medicine, Shanghai 201203, China
Abstract：Breast cancer and ovarian cancer are common malignant tumors in women, and the differences in their metabolic activity and protein structures reveal unique pathological mechanisms. However, due to the overlap in symptoms and molecular characteristics between the two, clinical diagnosis and differentiation remain challenging. A systematic study of the metabolic processes and protein conformational changes in breast cancer and ovarian cancer provides scientific evidence and guidance for disease diagnosis and personalized treatment. This study, based on attenuated total reflection Fourier transform infrared (ATR-FTIR) spectroscopy combined with machine learning methods, explores the differential spectral markers of breast cancer and ovarian cancer and evaluates their diagnostic and discriminative potential. A total of 157 female participants were included in the study, including 67 breast cancer patients, 41 ovarian cancer patients, and 49 healthy controls, with serum samples collected for spectral analysis. The results show that at the 1 450 cm-1 band, the absorbance of the breast cancer group was significantly higher than that of the ovarian cancer group (p<0.05), accompanied by a blue shift in the wavenumber, suggesting lipid metabolism and cell membrane synthesis abnormalities. Peak fitting analysis of the Amide I region revealed that the α-helix proportion in the breast cancer group was significantly lower than that in the ovarian cancer group (p<0.05). In comparisor the β-sheet proportion in the ovarian cancer group was significantly higher than that in the breast cancer group (p<0.05), revealing specific differences in protein conformation changes between the two cancers. The Linear Discriminant Analysis (LDA) model constructed using the relative intensity ratio of 1 450/1 650 cm-1 and Amide Ⅰ spectral data showed a reasonable differentiation performance (AUC=0.851, Specificity=73.2%, Sensitivity=80.3%). The results of this study indicate that ATR-FTIR spectroscopy combined with spectral feature analysis and classification models can provide effective support for the diagnosis and differentiation of breast cancer and ovarian cancer, laying the foundation for future cancer subtype diagnostic research.
Key words：ATR-FTIR spectroscopy; Serum; Breast cancer; Ovarian cancer; Linear discriminant analysis
宋 澳，蔡以撒，蔡李峥，杨万里，庞 楠，于瑞华，王士岩，杨 超，姜 峰. 基于ATR-FTIR光谱与机器学习的乳腺癌与卵巢癌差异性诊断研究[J]. 光谱学与光谱分析, 2025, 45(12): 3373-3380.
SONG Ao, CAI Yi-sa, CAI Li-zheng, YANG Wan-li, PANG Nan, YU Rui-hua, WANG Shi-yan, YANG Chao, JIANG Feng. Differential Diagnosis of Breast Cancer and Ovarian Cancer Based on ATR-FTIR Spectroscopy Coupled With Machine Learning. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(12): 3373-3380.
[1] Bray F, Laversanne M, Sung H, et al. CA: A Cancer Journal for Clinicians, 2024, 74(3): 229.
[2] LU Jin, WANG Jin-hua(卢 锦, 王金华). Cancer Research and Clinic(肿瘤研究与临床), 2018, 30(3): 204.
[3] LIANG You-feng, HAO Ming-xuan, GUO Rui, et al(梁有沣, 郝明炫, 郭 蕊, 等). Chinese Journal of Biotechnology(生物工程学报), 2023, 39(4): 1425.
[4] Baker M J, Trevisan J, Bassan P, et al. Nature Protocols, 2014, 9(8): 1771.
[5] Guo S, Wei G, Chen W, et al. Biomolecules, 2022, 12(12): 1815.
[6] Butler H J, Brennan P M, Cameron J M, et al. Nature Communications, 2019, 10(1): 4501.
[7] Akin Geyik G, Peker Cengiz B, Tugrul F, et al. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 2022, 267(Pt 2): 120619.
[8] Morais C L M, Lima K M G, Dickinson A W, et al. Analyst, 2024, 149(19): 4851.
[9] Zelig U, Barlev E, Bar O, et al. BMC Cancer, 2015, 15: 408.
[10] Lewis P D, Lewis K E, Ghosal R, et al. BMC Cancer, 2010, 10: 640.
[11] Francioso A, Punzi P, Boffi A, et al. Bioorganic & Medicinal Chemistry, 2015, 23(8): 1671.
[12] Hajian-Tilaki K. Caspian Journal of Internal Medicine, 2013, 4(2): 627.
[13] Gu Q, Li Z, Han J. Linear Discriminant Dimensionality Reduction. in: Machine Learning and Knowledge Discovery in Databases, Springer, 2011.
[14] Chen X, Chen C, Tian X, et al. Talanta, 2024, 266(2): 125052.
[15] Movasaghi Z, Rehman S, Rehman I U. Applied Spectroscopy Reviews, 2008, 43(2): 134.
[16] Gunasekaran S, Uthra D. Asian Journal of Chemistry, 2008, 20(7): 5695.
[17] Narayanan K, Thirunavukkarasu M, Jayanthi C E, et al. International Journal of Pharmacy and Biological Sciences, 2011, 1(2): 74.
[18] Yang S, Zhang Q, Yang H, et al. International Journal of Biological Macromolecules, 2022, 206: 175.
[19] Ali S, Naveed A, Hussain I, et al. Photodiagnosis and Photodynamic Therapy, 2023, 42: 103529.
[20] Tomas R C, Sayat A J, Atienza A N, et al. PLoS One, 2022, 17(1): e0262489.
[21] Ali L, Zhu C, Zhang Z, et al. IEEE Journal of Translational Engineering in Health and Medicine, 2019, 7: 2000410.
[22] Ricciardi V, Portaccio M, Lasalvia M, et al. Applied Sciences, 2022, 12: 5074.
[23] German M J, Hammiche A, Ragavan N, et al. Biophysical Journal, 2006, 90(10): 3783.
[24] Chen H, Li X, Zhang S, et al. Photodiagnosis and Photodynamic Therapy, 2022, 40: 103177.
[25] Yang X, Ou Q, Qian K, et al. Frontiers in Oncology, 2021, 11: 753791.
[26] Rakib F, Al-Saad K, Ahmed T, et al. Spectrochim Acta Part A: Molecular and Biomolecular Spectroscopy, 2021, 248: 119189.
[27] Byler D M, Susi H. Biopolymers, 1986, 25(3): 469.
[28] Jiang M, Shu T, Ye C, et al. Analyst, 2021, 146(8): 2490.
