INTRODUCTION Intraoperative diagnosis is essential for providing safe and effective care during cancer surgery. The existing workflow for intraoperative diagnosis based on H&E staining of processed tissue is time-, resource-, and labor-intensive. Moreover, interpretation of intraoperative histologic images is dependent on a pathology workforce that is contracting and unevenly distributed across the centers where cancer surgery is performed worldwide. METHODS We developed an automated workflow, independent of traditional H&E histology, that combines stimulated Raman histology (SRH), a rapid label-free optical imaging method, and deep convolutional neural networks (CNN) to predict diagnosis at the bedside in near real time. Specifically, our CNN, trained on over 2.5 million SRH images, predicts brain tumor diagnosis in the operating room in under 150 s, which is an order of magnitude faster than conventional techniques (eg, 20-30 min). RESULTS To validate our workflow in the clinical setting, we designed a multicenter, prospective, noninferiority clinical trial (N = 204) that compares SRH plus CNN vs traditional H&E histology. Primary endpoint was overall diagnostic accuracy. We show that CNN-based diagnosis of SRH images was noninferior to pathologist-based interpretation of conventional histologic images (overall accuracy, 94.6% vs 95.5%). Additionally, our CNN learned a hierarchy of interpretable histologic feature representations to classify the major histopathologic classes of brain tumors. We then developed and implemented a semantic segmentation method that can identify tumor infiltrated and diagnostic regions within SRH images. Mean intersection over union values was 61 ± 28.6 for ground truth diagnostic class and 86.0 ± 28.6 for tumor-infiltrated regions. CONCLUSION We have demonstrated how combining bedside optical histology with deep learning can result in near real-time intraoperative brain tumor diagnosis. Our workflow provides a means of delivering expert-level intraoperative diagnosis where neuropathology resources are scarce and improve diagnostic accuracy in resource-rich centers.

Mutant isocitrate-dehydrogenase 1 (mIDH1) synthesizes the oncometabolite 2-hydroxyglutarate (2HG), which elicits epigenetic reprogramming of the glioma cells' transcriptome by inhibiting DNA and histone demethylases. We show that the efficacy of immune-stimulatory gene therapy (TK/Flt3L) is enhanced in mIDH1 gliomas, due to the reprogramming of the myeloid cells' compartment infiltrating the tumor microenvironment (TME). We uncovered that the immature myeloid cells infiltrating the mIDH1 TME are mainly nonsuppressive neutrophils and preneutrophils. Myeloid cell reprogramming was triggered by granulocyte colony-stimulating factor (G-CSF) secreted by mIDH1 glioma stem/progenitor-like cells. Blocking G-CSF in mIDH1 glioma-bearing mice restores the inhibitory potential of the tumor-infiltrating myeloid cells, accelerating tumor progression. We demonstrate that G-CSF reprograms bone marrow granulopoiesis, resulting in noninhibitory myeloid cells within mIDH1 glioma TME and enhancing the efficacy of immune-stimulatory gene therapy.

Background Detection of glioma recurrence remains a challenge in modern neuro-oncology. Noninvasive radiographic imaging is unable to definitively differentiate true recurrence versus pseudoprogression. Even in biopsied tissue, it can be challenging to differentiate recurrent tumor and treatment effect. We hypothesized that intraoperative stimulated Raman histology (SRH) and deep neural networks can be used to improve the intraoperative detection of glioma recurrence. Methods We used fiber-laser-based SRH, a label-free, non-consumptive, high-resolution microscopy method (<60 secs per 1 x 1 mm2) to image a cohort of patients (n = 35) with suspected recurrent gliomas who underwent biopsy or resection. The SRH images were then used to train a convolutional neural network (CNN) and develop an inference algorithm to detect viable recurrent glioma. Following network training, the performance of the CNN was tested for diagnostic accuracy in a retrospective cohort (n = 48). Results Using patch-level CNN predictions, the inference algorithm returned a single Bernoulli distribution for the probability of tumor recurrence for each surgical specimen or patient. The external SRH validation dataset consisted of 48 patients (recurrent, 30; pseudoprogression, 18), and we achieved a diagnostic accuracy of 95.8%. Conclusion SRH with CNN-based diagnosis can be used to improve the intraoperative detection of glioma recurrence in near-real time. Our results provide insight into how optical imaging and computer vision can be combined to augment conventional diagnostic methods and improve the quality of specimen sampling at glioma recurrence.

Background Diagnostic delays impact the quality of life and survival of patients with brain tumors. Earlier and expeditious diagnoses in these patients are crucial to reducing the morbidities and mortalities associated with brain tumors. A simple, rapid blood test that can be administered easily in a primary care setting to efficiently identify symptomatic patients who are most likely to have a brain tumor would enable quicker referral to brain imaging for those who need it most. Methods Blood serum samples from 603 patients were prospectively collected and analyzed. Patients either had non-specific symptoms that could be indicative of a brain tumor on presentation to the Emergency Department, or a new brain tumor diagnosis and referral to the neurosurgical unit, NHS Lothian, Scotland. Patient blood serum samples were analyzed using the Dxcover®Brain Cancer liquid biopsy. This technology utilizes infrared spectroscopy combined with a diagnostic algorithm to predict the presence of intracranial disease. Results Our liquid biopsy approach reported an area under the receiver operating characteristic curve of 0.8. The sensitivity-tuned model achieves a 96% sensitivity with 45% specificity (NPV 99.3%) and identified 100% of glioblastoma multiforme patients. When tuned for a higher specificity, the model yields sensitivity of 47% with 90% specificity (PPV 28.4%). Conclusions This simple, non-invasive blood test facilitates the triage and radiographic diagnosis of brain tumor patients, while providing reassurance to healthy patients. Minimizing time to diagnosis would facilitate identification of brain tumor patients at an earlier stage, enabling more effective, less morbid surgical and adjuvant care.