Automated Volumetric Analysis of Postoperative Magnetic Resonance Imaging Predicts Survival in Patients with Glioblastoma.

BACKGROUND: Glioblastomas (GBMs) are primary brain tumors that are very difficult to treat. Magnetic resonance imaging (MRI) is the reference tool for diagnosis, postoperative control, and follow-up of GBM. The MRI tumor contrast enhancement part serves as a target for surgery. However, there are controversial data about the influence of pre- and postoperative tumor volumetric MRI parameters on overall survival (OS). METHODS: Data of 57 patients with GBM were analyzed retrospectively. All patients had maximum safe resection and standard adjuvant treatment. All patients underwent 1.5-T MRI with contrast in the first 24 hours postoperatively. The data of pre- and postoperative volumetric parameters were analyzed using the original software. RESULTS: Correlation analysis between the postoperative volume of the tumor contrast enhancement part and the patient's OS revealed a significant level (on the Chaddock scale) of inverse correlation. Residual tumor volume associated with OS of >6 months was determined as <2.5 cm3. The mortality risk in the first 6 months after tumor resection is 3.4 times higher when the tumor remnant is >2.5 cm3 (risk ratio, 3.4; P = 0.0002). CONCLUSIONS: The volume of MRI contrast-enhancing GBM remnants after surgery, automatically measured by the software, was a significant predictor for early postoperative progression and death.

New Software for Preoperative Diagnostics of Meningeal Tumor Histologic Types.

OBJECTIVE: Meningeal tumors are neoplasms with different histologic manifestations of both benign and malignant types that determine the prognosis of tumor recurrence and its consistency. The risk of surgical treatment depends on the location, size, and consistency of the tumor. Magnetic resonance imaging (MRI) sequences can be used to identify the features of tumors, but these MRI characteristics are not well understood. The present study describes an advanced mathematical algorithm to analyze MRI data and distinguish histologic types of meningeal tumors before surgery. METHODS: Forty-eight patients underwent surgical removal of meningeal brain tumor. All patients had preoperative MRI with a 1.5-T scanner. One radiologist and 2 neurosurgeons evaluated MRI histogram peaks of the whole tumor volume using the advanced computer algorithm. RESULTS: Three specialists received the following mean value of histogram peaks: 15.99 ± 0.23 (± standard error of the mean [SEM]) for meningoteliomatous meningiomas; 21.24 ± 0.3 (±SEM) for fibroplastic meningiomas; 19.0 ± 0.28 (±SEM) for transitional meningiomas; 10.7 ± 0.27 (±SEM) for anatypical, anaplastic meningiomas, 11.03 ± 0.51 (±SEM) for primary intracranial fibrosarcomas and 25.72 ± 0.29 (±SEM) for meningeal hemangiopericytomas. A one-way analysis of variance test proved the difference between group means: F = 70.138, P < 0.01. The Tukey test and the Games-Howell test indicated that the difference between the tumor groups was significant. Mean deviation in agreement index between specialists was 0.98 ± 0.007 (±SEM). CONCLUSIONS: The advanced algorithm proved high specificity, sensitivity, and interoperator repeatability.