Rapid intraoperative diagnosis of pediatric brain tumors using Raman spectroscopy: A machine learning approach

Abstract

Background Surgical resection is a mainstay in the treatment of pediatric brain tumors to achieve tissue diagnosis and tumor debulking. While maximal safe resection of tumors is desired, it can be challenging to differentiate normal brain from neoplastic tissue using only microscopic visualization, intraoperative navigation, and tactile feedback. Here, we investigate the potential for Raman spectroscopy (RS) to accurately diagnose pediatric brain tumors intraoperatively. Methods Using a rapid acquisition RS device, we intraoperatively imaged fresh ex vivo brain tissue samples from 29 pediatric patients at the Lucile Packard Children’s Hospital between October 2018 and March 2020 in a prospective fashion. Small tissue samples measuring 2-4 mm per dimension were obtained with each individual tissue sample undergoing multiple unique Raman spectra acquisitions. All tissue samples from which Raman spectra were acquired underwent individual histopathology review. A labeled dataset of 684 unique Raman spectra gathered from 156 samples was then used to develop a machine learning model capable of 1) differentiating normal brain from tumor tissue and 2) normal brain from low grade glioma (LGG) tissue. Results Trained logistic regression model classifiers were developed using our labeled dataset. Model performance was evaluated using leave-one-patient-out cross-validation. The area-under the-curve (AUC) of the receiver-operating-characteristic (ROC) curve for our tumor versus normal brain model was 0.94. The AUC of the ROC curve for LGG versus normal brain was 0.91. Conclusions Our work suggests that RS can be used to develop a machine learning based classifier to differentiate tumor versus non-tumor tissue during resection of pediatric brain tumors.

Raman spectroscopy: A prospective intraoperative visualization technique for gliomas

“…One of the most common and direct applications of Raman spectroscopy is to examine the intraoperatively obtained tissue directly, and the results derived in this way represent the actual intraoperative situation of the tissue more directly and better than FFPE samples. A prospective study in 2022 carried out direct Raman imaging on 29 freshly collected ex vivo brain tissue samples, each of which was split into 2-4 mm and examined for pathology, and finally, the results were trained by machine learning algorithms as a predictive model that could classify tumors from normal tissue with an accuracy of 89.8%, sensitivity of 84.9%, and specificity of 92.3%, as well as LGG and normal tissue with an accuracy of 86.2%, sensitivity of 91.3%, and specificity of 81.2% (Figure 2) (30). By analyzing a total of 3450 spectral results from 63 fresh glioma samples, Riva et al identified 19 Raman shifts specific to glioma, whereby the predictions could reach 82% precision (28).…”

“…Subsequent studies have gradually overcome the low signal-to-noise ratio and low sensitivity of the earlier technique in recent years with the development of optical and data processing techniques, and studies using the technique to identify tumor cells on postoperative Formalin-fixed paraffin-embedded (FFPE) pathology sections have gradually emerged ( 15 , 27 ). Latest studies have also utilized intraoperative frozen pathology sections ( 14 ), intraoperative fresh tissue blocks ( 28 ), and in situ tissue in the operative area ( 11 , 29 ) for the identification and diagnosis of gliomas, while data processing techniques such as machine learning and deep learning have also been used to enhance the accuracy of Raman techniques ( 30 – 32 ).…”

“…(C) ROC curve of a trained logistic regression model to identify tumor and normal brain tissue. (D) ROC curve of a trained logistic regression model to identify low grade glioma and normal brain tissue ( 30 ).…”

Raman Spectroscopy: A Tool for Molecular Fingerprinting of Brain Cancer

“…The accuracy was achieved at around 85% in both cases. 30 A label-free in situ intraoperative cancer detection system based on high wavenumber (2000 to 4000 cm −1 ) Raman spectroscopy was developed by Leblond and team for the detection and biopsy of brain cancer. They validated the integrated core needle biopsy system with an animal study on swine.…”

Artificial intelligence applications in pediatric oncology diagnosis

“…Khayat Kashani et al [45] used inflammatory indicators on peripheral blood tests to classify pediatric benign and malignant brain tumors. Jabarkheel et al [46] explored the intraoperative diagnostic potential of Raman spectroscopy using an ML classifier in pediatric brain tumors. This ML-based method differentiated the normal brain from neoplastic tissue in a non-invasive manner and efficient diagnostic performance (AUC > 0.90) compared with microscopic visualization and intraoperative navigation.…”