A Deep Learning Model for Preoperative Differentiation of Glioblastoma, Brain Metastasis and Primary Central Nervous System Lymphoma: A Pilot Study

Abstract

BackgroundNeuroimaging differentiation of glioblastoma, primary central nervous system lymphoma (PCNSL) and solitary brain metastasis (BM) remains challenging in specific cases showing similar appearances or atypical features. Overall, advanced MRI protocols have high diagnostic reliability, but their limited worldwide availability, coupled with the overlapping of specific neuroimaging features among tumor subgroups, represent significant drawbacks and entail disparities in the planning and management of these oncological patients.ObjectiveTo evaluate the classification performance metrics of a deep learning algorithm trained on T1-weighted gadolinium-enhanced (T1Gd) MRI scans of glioblastomas, atypical PCNSLs and BMs.Materials and MethodsWe enrolled 121 patients (glioblastoma: n=47; PCNSL: n=37; BM: n=37) who had undergone preoperative T1Gd-MRI and histopathological confirmation. Each lesion was segmented, and all ROIs were exported in a DICOM dataset. The patient cohort was then split in a training and hold-out test sets following a 70/30 ratio. A Resnet101 model, a deep neural network (DNN), was trained on the training set and validated on the hold-out test set to differentiate glioblastomas, PCNSLs and BMs on T1Gd-MRI scans.ResultsThe DNN achieved optimal classification performance in distinguishing PCNSLs (AUC: 0.98; 95%CI: 0.95 - 1.00) and glioblastomas (AUC: 0.90; 95%CI: 0.81 - 0.97) and moderate ability in differentiating BMs (AUC: 0.81; 95%CI: 0.70 - 0.95). This performance may allow clinicians to correctly identify patients eligible for lesion biopsy or surgical resection.ConclusionWe trained and internally validated a deep learning model able to reliably differentiate ambiguous cases of PCNSLs, glioblastoma and BMs by means of T1Gd-MRI. The proposed predictive model may provide a low-cost, easily-accessible and high-speed decision-making support for eligibility to diagnostic brain biopsy or maximal tumor resection in atypical cases.

Convolutional Neural Network Techniques for Brain Tumor Classification (from 2015 to 2022): Review, Challenges, and Future Perspectives

“…Normal vs. LGG vs. HGG Custom CNN model 5-fold CV 95 2. MEN vs. Glioma vs. PT Custom CNN model 5-fold CV 94.74 Guo et al [ 128 ] 2022 GBM vs. AST vs. OLI Custom CNN 3-fold CV SEN = 0.772, SPE = 93.0%, AUC = 0.902 87.8% Rizwan et al [ 130 ] 2022 Grade I vs. Grade II vs. Grade III Custom CNN No info shared 97.14% Tariciotti et al [ 123 ] 2022 Metastasis vs. GBM vs. PCNSL Resnet101 Hold-out PRE = 91.88%, SEN = 90.84%, SPE = 96.34%, F1 score = 91.0%, AUC = 0.92 94.72% 4 classes Ahammed et al [ 72 ] 2019 Grade I vs. Grade II vs. Grade III vs. Grade IV VGG19 No info shared PRE = 94.71%, SEN = 92.72%, SPE = 98.13%, F1 score = 93.71% 98.25 Mohammed et al [ 51 ] 2020 …”

The Role of [68Ga]Ga-DOTA-SSTR PET Radiotracers in Brain Tumors: A Systematic Review of the Literature and Ongoing Clinical Trials

“…Although multiparametric MRI represents the current imaging gold standard in primary and metastatic brain neoplasms, PET studies have the unique and complementary ability to evaluate and characterize the metabolic patterns within the tumor and non-tumor tissues through the use of selected radiolabeled tracers [ 84 , 85 , 86 ]. The roles of different PET tracers in tumor diagnosis and post-treatment response assessment have been largely discussed and validated by international consensuses and recommendations [ 87 , 88 , 89 , 90 ].…”

Use of 18F-FDG-PET/CT in differential diagnosis of primary central nervous system lymphoma and high-grade gliomas: A meta-analysis

“…Another meta-analysis of 10 articles by You et al suggested that arterial spin labeling could differentiate between PCNSL and HGG with an AUC of 0.86 ( 27 ). Additionally, a diagnostic model constructed with a deep learning algorithm based on MR imaging was used to differentiate PCNSLs, glioblastoma and brain metastases, with an AUC of 0.98 in the differential diagnosis of PCNSL ( 28 ). However, PET-CT still has great diagnostic advantages over other techniques in terms of imaging detection methods that rely on a single index.…”