Machine Learning-Based Fluorescence Assessment for Augmented Imaging and Decision Support in Glioblastoma Resections. Schaufler, A., Stein, K., Pamnani, S., Dumitru, C. A., Neyazi, B., Rashidi, A., Boese, A., & Sandalcioglu, I. E. Cancers, 18(7):1125, March, 2026. ![Machine Learning-Based Fluorescence Assessment for Augmented Imaging and Decision Support in Glioblastoma Resections [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Background/Objectives: Glioblastoma is the most common and aggressive primary malignant brain tumor in adults, characterized by infiltrative growth and poor prognosis. Achieving maximal resection without inducing neurological deficits remains a challenge in glioblastoma surgery. While 5-aminolevulinic acid-based fluorescence-guided surgery supports intraoperative tumor visualization, its reliability is limited by patient variability and weak fluorescence signals. This study proposes a machine learning framework to enhance fluorescence-guided surgery sensitivity by analyzing surgical microscope images at the pixel level. Methods: Fluorescence-mode neurosurgical microscope images of synthetic samples with known Protoporphyrin IX (PPIX) concentrations were used to train three classifiers (Support Vector Machine, Naïve Bayes, Neural Network) for pixel-wise fluorescence detection. In parallel, three contrastive-learning-based Variational Autoencoders (VAE, β = 1, 2, 3) were evaluated for detecting weak fluorescence beyond visual perception. Additionally, a regression model was trained to relate pixel features to PPIX concentration. The best-performing VAE (β = 1) was subsequently trained on real intraoperative data, and its detection sensitivity was compared to annotations from four experienced surgeons. Results: The proposed model achieved the highest detection rates on synthetic test data when calibrated for 99% specificity. Applied to real intraoperative images, the model revealed fluorescent areas substantially larger than those marked by experienced surgeons. In non-5-ALA control cases, minimal false positives were observed, indicating a specificity exceeding 99.9%. The regression model reliably quantified PPIX concentration in synthetic samples (R2=0.92). Conclusions: By enabling more sensitive and objective fluorescence detection, this approach offers a valuable tool for improving surgical decision-making and facilitating safer, more extensive tumor resections.
@article{schaufler_machine_2026,
title = {Machine {Learning}-{Based} {Fluorescence} {Assessment} for {Augmented} {Imaging} and {Decision} {Support} in {Glioblastoma} {Resections}},
volume = {18},
issn = {2072-6694},
url = {https://www.mdpi.com/2072-6694/18/7/1125},
doi = {10.3390/cancers18071125},
abstract = {Background/Objectives: Glioblastoma is the most common and aggressive primary malignant brain tumor in adults, characterized by infiltrative growth and poor prognosis. Achieving maximal resection without inducing neurological deficits remains a challenge in glioblastoma surgery. While 5-aminolevulinic acid-based fluorescence-guided surgery supports intraoperative tumor visualization, its reliability is limited by patient variability and weak fluorescence signals. This study proposes a machine learning framework to enhance fluorescence-guided surgery sensitivity by analyzing surgical microscope images at the pixel level. Methods: Fluorescence-mode neurosurgical microscope images of synthetic samples with known Protoporphyrin IX (PPIX) concentrations were used to train three classifiers (Support Vector Machine, Naïve Bayes, Neural Network) for pixel-wise fluorescence detection. In parallel, three contrastive-learning-based Variational Autoencoders (VAE, β = 1, 2, 3) were evaluated for detecting weak fluorescence beyond visual perception. Additionally, a regression model was trained to relate pixel features to PPIX concentration. The best-performing VAE (β = 1) was subsequently trained on real intraoperative data, and its detection sensitivity was compared to annotations from four experienced surgeons. Results: The proposed model achieved the highest detection rates on synthetic test data when calibrated for 99\% specificity. Applied to real intraoperative images, the model revealed fluorescent areas substantially larger than those marked by experienced surgeons. In non-5-ALA control cases, minimal false positives were observed, indicating a specificity exceeding 99.9\%. The regression model reliably quantified PPIX concentration in synthetic samples (R2=0.92). Conclusions: By enabling more sensitive and objective fluorescence detection, this approach offers a valuable tool for improving surgical decision-making and facilitating safer, more extensive tumor resections.},
language = {en},
number = {7},
urldate = {2026-04-01},
journal = {Cancers},
author = {Schaufler, Anna and Stein, Klaus-Peter and Pamnani, Sunisha and Dumitru, Claudia A. and Neyazi, Belal and Rashidi, Ali and Boese, Axel and Sandalcioglu, I. Erol},
month = mar,
year = {2026},
pages = {1125},
}
Downloads: 0
{"_id":"RpGNEygK9aXxFFkSD","bibbaseid":"schaufler-stein-pamnani-dumitru-neyazi-rashidi-boese-sandalcioglu-machinelearningbasedfluorescenceassessmentforaugmentedimaginganddecisionsupportinglioblastomaresections-2026","author_short":["Schaufler, A.","Stein, K.","Pamnani, S.","Dumitru, C. A.","Neyazi, B.","Rashidi, A.","Boese, A.","Sandalcioglu, I. E."],"bibdata":{"bibtype":"article","type":"article","title":"Machine Learning-Based Fluorescence Assessment for Augmented Imaging and Decision Support in Glioblastoma Resections","volume":"18","issn":"2072-6694","url":"https://www.mdpi.com/2072-6694/18/7/1125","doi":"10.3390/cancers18071125","abstract":"Background/Objectives: Glioblastoma is the most common and aggressive primary malignant brain tumor in adults, characterized by infiltrative growth and poor prognosis. Achieving maximal resection without inducing neurological deficits remains a challenge in glioblastoma surgery. While 5-aminolevulinic acid-based fluorescence-guided surgery supports intraoperative tumor visualization, its reliability is limited by patient variability and weak fluorescence signals. This study proposes a machine learning framework to enhance fluorescence-guided surgery sensitivity by analyzing surgical microscope images at the pixel level. Methods: Fluorescence-mode neurosurgical microscope images of synthetic samples with known Protoporphyrin IX (PPIX) concentrations were used to train three classifiers (Support Vector Machine, Naïve Bayes, Neural Network) for pixel-wise fluorescence detection. In parallel, three contrastive-learning-based Variational Autoencoders (VAE, β = 1, 2, 3) were evaluated for detecting weak fluorescence beyond visual perception. Additionally, a regression model was trained to relate pixel features to PPIX concentration. The best-performing VAE (β = 1) was subsequently trained on real intraoperative data, and its detection sensitivity was compared to annotations from four experienced surgeons. Results: The proposed model achieved the highest detection rates on synthetic test data when calibrated for 99% specificity. Applied to real intraoperative images, the model revealed fluorescent areas substantially larger than those marked by experienced surgeons. In non-5-ALA control cases, minimal false positives were observed, indicating a specificity exceeding 99.9%. The regression model reliably quantified PPIX concentration in synthetic samples (R2=0.92). Conclusions: By enabling more sensitive and objective fluorescence detection, this approach offers a valuable tool for improving surgical decision-making and facilitating safer, more extensive tumor resections.","language":"en","number":"7","urldate":"2026-04-01","journal":"Cancers","author":[{"propositions":[],"lastnames":["Schaufler"],"firstnames":["Anna"],"suffixes":[]},{"propositions":[],"lastnames":["Stein"],"firstnames":["Klaus-Peter"],"suffixes":[]},{"propositions":[],"lastnames":["Pamnani"],"firstnames":["Sunisha"],"suffixes":[]},{"propositions":[],"lastnames":["Dumitru"],"firstnames":["Claudia","A."],"suffixes":[]},{"propositions":[],"lastnames":["Neyazi"],"firstnames":["Belal"],"suffixes":[]},{"propositions":[],"lastnames":["Rashidi"],"firstnames":["Ali"],"suffixes":[]},{"propositions":[],"lastnames":["Boese"],"firstnames":["Axel"],"suffixes":[]},{"propositions":[],"lastnames":["Sandalcioglu"],"firstnames":["I.","Erol"],"suffixes":[]}],"month":"March","year":"2026","pages":"1125","bibtex":"@article{schaufler_machine_2026,\n\ttitle = {Machine {Learning}-{Based} {Fluorescence} {Assessment} for {Augmented} {Imaging} and {Decision} {Support} in {Glioblastoma} {Resections}},\n\tvolume = {18},\n\tissn = {2072-6694},\n\turl = {https://www.mdpi.com/2072-6694/18/7/1125},\n\tdoi = {10.3390/cancers18071125},\n\tabstract = {Background/Objectives: Glioblastoma is the most common and aggressive primary malignant brain tumor in adults, characterized by infiltrative growth and poor prognosis. Achieving maximal resection without inducing neurological deficits remains a challenge in glioblastoma surgery. While 5-aminolevulinic acid-based fluorescence-guided surgery supports intraoperative tumor visualization, its reliability is limited by patient variability and weak fluorescence signals. This study proposes a machine learning framework to enhance fluorescence-guided surgery sensitivity by analyzing surgical microscope images at the pixel level. Methods: Fluorescence-mode neurosurgical microscope images of synthetic samples with known Protoporphyrin IX (PPIX) concentrations were used to train three classifiers (Support Vector Machine, Naïve Bayes, Neural Network) for pixel-wise fluorescence detection. In parallel, three contrastive-learning-based Variational Autoencoders (VAE, β = 1, 2, 3) were evaluated for detecting weak fluorescence beyond visual perception. Additionally, a regression model was trained to relate pixel features to PPIX concentration. The best-performing VAE (β = 1) was subsequently trained on real intraoperative data, and its detection sensitivity was compared to annotations from four experienced surgeons. Results: The proposed model achieved the highest detection rates on synthetic test data when calibrated for 99\\% specificity. Applied to real intraoperative images, the model revealed fluorescent areas substantially larger than those marked by experienced surgeons. In non-5-ALA control cases, minimal false positives were observed, indicating a specificity exceeding 99.9\\%. The regression model reliably quantified PPIX concentration in synthetic samples (R2=0.92). Conclusions: By enabling more sensitive and objective fluorescence detection, this approach offers a valuable tool for improving surgical decision-making and facilitating safer, more extensive tumor resections.},\n\tlanguage = {en},\n\tnumber = {7},\n\turldate = {2026-04-01},\n\tjournal = {Cancers},\n\tauthor = {Schaufler, Anna and Stein, Klaus-Peter and Pamnani, Sunisha and Dumitru, Claudia A. and Neyazi, Belal and Rashidi, Ali and Boese, Axel and Sandalcioglu, I. Erol},\n\tmonth = mar,\n\tyear = {2026},\n\tpages = {1125},\n}\n\n\n\n","author_short":["Schaufler, A.","Stein, K.","Pamnani, S.","Dumitru, C. A.","Neyazi, B.","Rashidi, A.","Boese, A.","Sandalcioglu, I. E."],"key":"schaufler_machine_2026","id":"schaufler_machine_2026","bibbaseid":"schaufler-stein-pamnani-dumitru-neyazi-rashidi-boese-sandalcioglu-machinelearningbasedfluorescenceassessmentforaugmentedimaginganddecisionsupportinglioblastomaresections-2026","role":"author","urls":{"Paper":"https://www.mdpi.com/2072-6694/18/7/1125"},"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://bibbase.org/zotero/ABOESE","dataSources":["m2XiccSnY5G7F7fcM"],"keywords":[],"search_terms":["machine","learning","based","fluorescence","assessment","augmented","imaging","decision","support","glioblastoma","resections","schaufler","stein","pamnani","dumitru","neyazi","rashidi","boese","sandalcioglu"],"title":"Machine Learning-Based Fluorescence Assessment for Augmented Imaging and Decision Support in Glioblastoma Resections","year":2026}