AI Spectroscopic Analysis
Collaboration with Leading Lithuanian Hospital / Case
AAI Labs partnered with a leading Lithuanian hospital Santaros Klinikos to tackle a critical issue in nephrology: the slow and inaccurate analysis of kidney stones.
In Lithuania, an estimated 3,000 kidney stone removal surgeries are conducted annually, and the need for precise analysis of the stones’ composition is essential. Misidentification of kidney stones often leads to ineffective treatments, recurrent stone formations, and higher risks of complications, including kidney infections and renal failure.
Traditional method of manual chemical analysis for stone composition is time-consuming and prone to human error. Even with Fourier-transform infrared (FTIR) spectroscopy medical professionals often face delays due to the manual interpretation required, leading to increased risks of misdiagnosis or improper treatment plans. Santaros Klinikos sought to streamline this process to provide faster, more reliable diagnoses, reduce patient readmissions, and plan more effective treatment paths.
"The created prototype and prospective further development of this technology will undoubtedly improve the quality of patient treatment and will allow Santaros Klinikos to be an innovator in the application of ML methods in Lithuania," said Tadas Šubonis, CTO of AAI Labs.
Solution
AAI Labs introduced an AI-powered FTIR spectral analysis system to improve the process of identifying kidney stone compositions. Utilizing AI's ability to rapidly analyze vast amounts of spectroscopic data, the system compares the spectra of a patient’s kidney stone (black line) to a library of possible material compositions (gray lines). This innovation enables doctors to immediately identify key elements of kidney stones and predict their formation more accurately.
By automating the interpretation process through advanced AI algorithms, the system can reduce analysis time from hours to minutes. The AI models were trained using hundreds of existing spectra samples to recognize the unique signatures of different stone types, such as calcium oxalate dihydrate (blue line), uric acid stones, and others. The AI engine assigns likelihood scores for potential matches, allowing doctors to base treatment decisions on highly accurate, data-driven insights.
The solution offers several benefits:
Speed - the AI system reduced the time taken for analysis by up to 80%, providing almost instantaneous results to guide treatment.
Accuracy - machine learning algorithms ensure greater precision, reducing the risk of human error that can occur during manual analysis.
Patient outcomes - with faster diagnoses, doctors can implement tailored treatment plans that reduce the likelihood of recurrence, which improves patient care and minimizes the long-term costs of treatment.
Potential applications to other fields
While this AI-powered FTIR system was initially developed for kidney stone analysis, its applications extend far beyond nephrology. AI-enhanced spectroscopic analysis can be adapted for use in other areas of medicine, research, and industrial applications.
Oncology: cancer diagnosis
FTIR spectroscopy can be applied to tissue sample analysis to differentiate between benign and malignant tumors. With AI, the system could analyze cancerous tissue samples and identify the type and stage of cancer more precisely, offering a less invasive and quicker diagnostic method than current biopsy techniques. As Kumari et al. in their 2018 paper argue, AI-driven FTIR could enhance early cancer detection, particularly for difficult-to-diagnose cancers like pancreatic and ovarian cancer.
Pharmaceuticals: drug purity testing
In the pharmaceutical industry, ensuring the purity and quality of drugs is vital for patient safety. AI-enhanced FTIR could be applied to inspect drug compositions, rapidly identifying impurities or contaminants. For example, during the production of antibiotics, the AI model could ensure consistent purity, reducing risks to patients while improving manufacturing efficiency.
Food and agriculture: quality control
AI-driven spectroscopic analysis can be applied to agricultural products to ensure the quality and safety of food supplies. By analyzing the chemical composition of crops, fruits, and vegetables, the system can detect contamination, pesticide residue, and other harmful substances. This could improve quality control processes in industries like winemaking and dairy production, where spectral analysis is already used, but at a slower, manual pace.
Environmental monitoring
In environmental sciences, AI-enhanced spectroscopic analysis can be used for real-time air and water quality monitoring. By identifying pollutants, such as heavy metals or volatile organic compounds (VOCs), it can offer timely insights for environmental protection agencies and industrial regulators to prevent harm to ecosystems and public health. This technology can be invaluable in identifying oil spills, chemical leaks, or air pollution levels more accurately and efficiently than existing technologies.
Client
Vilnius University Hospital Santaros klinikos