Stroke Mimics and Chameleons from the Radiological Viewpoint of Glioma Diagnosis

Robust Ensemble of Two Different Multimodal Approaches to Segment 3D Ischemic Stroke Segmentation Using Brain Tumor Representation Among Multiple Center Datasets

Ischemic stroke segmentation at an acute stage is vital in assessing the severity of patients' impairment and guiding therapeutic decision-making for reperfusion. Although many deep learning studies have shown attractive performance in medical segmentation, it is difficult to use these models trained on public data with private hospitals' datasets. Here, we demonstrate an ensemble model that employs two different multimodal approaches for generalization, a more effective way to perform on external datasets. First, after we jointly train a segmentation model on diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) MR modalities, the model is inferred on the DWI images. Second, a channel-wise segmentation model is trained by concatenating the DWI and ADC images as input, and then is inferred using both MR modalities. Before training with ischemic stroke data, we utilized BraTS 2021, a public brain tumor dataset, for transfer learning. An extensive ablation study evaluates which strategy learns better representations for ischemic stroke segmentation. In our study, nnU-Net well-known for robustness is selected as our baseline model. Our proposed method is evaluated on three different datasets: the Asan Medical Center (AMC) I and II, and the 2022 Ischemic Stroke Lesion Segmentation (ISLES). Our experiments are widely validated over a large, multi-center, and multi-scanner dataset with a huge amount of 846 scans. Not only stroke lesion models can benefit from transfer learning using brain tumor data, but combining the MR modalities using different training schemes also highly improves segmentation performance. The method achieved a top-1 ranking in the ongoing ISLES'22 challenge and performed particularly well on lesion-wise metrics of interest to neuroradiologists, achieving a Dice coefficient of 78.69% and a lesion-wise F1 score of 82.46%. Also, the method was relatively robust on the AMC I (Dice, 60.35%; lesion-wise F1, 68.30%) and II (Dice; 74.12%; lesion-wise F1, 67.53%) datasets in different settings. The high segmentation accuracy of our proposed method could improve radiologists' ability to detect ischemic stroke lesions in MRI images. Our model weights and inference code are available on https://github.com/MDOpx/ISLES22-model-inference .

Dural Arteriovenous Fistula Mimicking a Brain Tumor on Methionine-positron Emission Tomography: A Case Report

In this article, we report a case wherein a brain tumor was suspected based on computed tomography and magnetic resonance imaging findings. We made an initial diagnosis of malignant brain tumor based on methionine-positron emission tomography (PET) findings, but the correct diagnosis was dural arteriovenous fistula (DAVF). The patient was a 45-year-old man with DAVF who developed headache. Methionine-PET imaging showed high methionine uptake in the lesion. Although the tumor was strongly suspected from the findings of methionine-PET, the diagnosis of DAVF could be made correctly only by interpreting digital subtraction angiography and computed tomographic angiography. The findings of methionine-PET, which is considered useful in the diagnosis and denial of brain tumors, made the diagnosis of DAVF more difficult. The increased uptake of methionine-PET in DAVF is an important finding because, to our knowledge, this study is the first to report such finding. The results of this study might be useful for differential diagnoses when the diagnosis is uncertain.

The landscape of brain tumor mimics in neuro-oncology practice

BACKGROUND AND OBJECTIVE

Differentiating neoplastic and non-neoplastic brain lesions is essential to make management recommendations and convey prognosis, but the distinction between brain tumors and their mimics in practice may prove challenging. The aim of this study is to provide the incidence of brain tumor mimics in the neuro-oncology setting and describe this patient subset.

METHODS

Retrospective study of adult patients referred to the Division of Neuro-oncology for a presumed diagnosis of brain tumor from January 1, 2005 through December 31, 2017, who later satisfied the diagnosis of a non-neoplastic entity based on neuroimaging, clinical course, and/or histopathology evaluation. We classified tumor mimic entities according to clinical, radiologic, and laboratory characteristics that correlated with the diagnosis.

RESULTS

The incidence of brain tumor mimics was 3.4% (132/3897). The etiologies of the non-neoplastic entities were vascular (35%), inflammatory non-demyelinating (26%), demyelinating (15%), cysts (10%), infectious (9%), and miscellaneous (5%). In our study, 38% of patients underwent biopsy to determine diagnosis, but in 26%, the biopsy was inconclusive.

DISCUSSION

Brain tumor mimics represent a small but important subset of the neuro-oncology referrals. Vascular, inflammatory, and demyelinating etiologies represent two-thirds of cases. Recognizing the clinical, radiologic and laboratory characteristics of such entities may improve resource utilization and prevent unnecessary as well as potentially harmful diagnostic and therapeutic interventions.