Gastric cancer poses a significant health concern, as evidenced by the annual diagnosis of approximately 990,000 individuals worldwide and the subsequent occurrence of approximately 738,000 fatalities [1]. It ranks as the fifth most prevalent form of cancer and stands as the fourth primary contributor to cancer-related mortality [2]. The etiology of Gastric cancer encompasses a range of risk factors, including both environmental and genetic influences, with Helicobacter pylori infection accounting for approximately 90% of distal Gastric cancer cases [3,4,5]. Additionally, excessive body fat, tobacco use, and the consumption of salted and processed meats have been identified as other notable risk factors [6].

The identification of gastric cancer encompasses various diagnostic methods, including endoscopy, blood tests, and imaging techniques [7]. While endoscopy is widely regarded as the preferred diagnostic tool [8], it is also the most invasive and carries the risk of complications such as bleeding or perforation. Blood tests are employed to detect specific proteins generated by cells within the body, including cancer cells. These proteins, known as tumor markers, are typically utilized in conjunction with other diagnostic approaches. The commonly employed tumor markers for gastric cancer include carcinoembryonic antigen (CEA) and carbohydrate antigen, such as CA19-9 or CA72-41 [9, 10]. Nevertheless, these markers exhibit limitations and a lack of specificity in identifying precancerous lesions associated with gastric cancer, such as atrophic gastritis, intestinal metaplasia, and dysplasia. In addition to tumor markers, imaging modalities such as X-ray imaging, computed tomography (CT) scan, and magnetic resonance imaging (MRI) can also be utilized for the detection of gastric cancer [11, 12]. These technologies have the capability to offer insights into tumor size and location, yet their accuracy falls short compared to endoscopy. Consequently, international research teams are actively seeking novel diagnostic approaches, among which ATR-FTIR spectroscopy emerges as a potentially valuable option due to its high sensitivity, non-destructive nature, and relatively affordable cost [13, 14].

ATR-FTIR spectroscopy has emerged as a promising technique for cancer diagnostic testing, enabling the measurement of the fingerprint spectra of compounds at the molecular level [15]. This technique capitalizes on the significant changes or conformational alterations in the bonding and vibrational patterns of functional groups within biomolecules, such as nucleic acids, proteins, lipids, and carbohydrates, that occur during cancer development. Consequently, ATR-FTIR spectroscopy can effectively differentiate between normal and tumor tissues [16]. Gonul et al. identified significant disparities between tissue samples from cancer patients and those from healthy individuals using ATR-FTIR spectroscopy. They further demonstrated that spectral parameters, including lipid/protein concentration and nucleic acid/protein concentration, could serve as biomarkers for the early diagnosis of cancer [17]. Additionally, Maiti et al. employed ultra-wideband mid-infrared Fourier absorption spectroscopy to monitor the respiration of volunteers, achieving an accuracy exceeding 95% across four spectral ranges in differentiating healthy individuals from prostate cancer patients [18]. However, the samples utilized in these studies exhibit certain limitations. Gastric tissue samples are challenging to procure and pose potential harm to patients. Exhalation samples are intricate to analyze and vulnerable to environmental interferences. In contrast, serum samples offer several advantages: they are straightforward to collect and store, and they exhibit reduced susceptibility to environmental disturbances. Furthermore, serum comprises water, organic compounds, and inorganic salts, making it a reliable medium for reflecting the physiological and pathological states of the human body. Sheng et al. demonstrated that the RNA/DNA ratio in the serum of gastric cancer patients (n = 27) was significantly lower compared to that of healthy individuals (n = 19). Furthermore, the ratio of H₂₉₉₅₉/H₂₉₃₁ was found to be an effective biomarker for distinguishing between the serum of gastric cancer patients and that of healthy controls [19]. Recently, Guo et al. utilized ATR-FTIR spectroscopy to analyze blood samples from 68 gastric cancer patients, 73 colorectal cancer patients, 25 liver cancer patients, and 44 healthy individuals. They employed machine learning classification techniques and, by analyzing the 2-Dimensional Second Derivative Infrared Spectroscopy, found that the Back Propagation algorithm had the highest accuracy, exceeding 95% [20]. These findings indicate that ATR-FTIR spectroscopy holds substantial promise for the diagnosis of gastric cancer. Future research should focus on increasing sample sizes and incorporating a broader range of machine learning methodologies to enhance diagnostic accuracy.

The present study demonstrates the feasibility of utilizing serum-based ATR-FTIR spectroscopy for the detection of gastric cancer. Employing PCA and machine learning algorithms, we effectively analyzed 192 human serum samples to differentiate between gastric cancer patients and healthy controls. This approach offers a viable and non-invasive method for the clinical screening and diagnosis of gastric cancer.