Tumor Core Research Articles

HybridCSF model for magnetic resonance image based brain tumor segmentation

<p>The human brain comprises a complex interconnection of nerve cells and vital organs, which regulates crucial bodily processes. Although neurons commonly undergo developmental stages, they may occasionally experience abnormalities, leading to abnormal growths known as brain tumors. The objective of brain tumor segmentation is to produce precise boundaries of brain tumor regions. This study extensively analyzes deep learning methods for brain tumor detection, evaluating their effectiveness across diverse datasets. It introduces a hybrid model, which is proposed by the name HybriCSF: hybrid convolutional-SVM-fuzzy C-means model combining convolutional neural network (CNN) with the classifier support vector machine (SVM) and clustering technique fuzzy C-means (FCM). The proposed model was implemented on Br35H, BraTs 2020 and BraTs2021 datasets. The suggested model outperformed the existing methods by achieving 98.6% of accuracy on Br35H dataset and dice score of 0.63, 0.87, 0.81 on BraTs 2020 dataset for enhancing tumor (ET), whole tumor (WT), and tumor core (TC), respectively. The achieved dice scores on the BraTs 2021datasets are 0.89, 0.95, and 0.89 for ET, WT, and TC, respectively. The results show that the suggested model HybriCSF outperforms the other CNN-based models in terms of accuracy.</p>

NnU-Net-based Segmentation of Tumor Subcompartments in Pediatric Medulloblastoma Using Multiparametric MRI: A Multi-institutional Study.

Purpose To evaluate nnU-Net-based segmentation models for automated delineation of medulloblastoma tumors on multi-institutional MRI scans. Materials and Methods This retrospective study included 78 pediatric patients (52 male, 26 female), with ages ranging from 2 to 18 years, with medulloblastomas, from three different sites (28 from hospital A, 18 from hospital B, and 32 from hospital C), who had data available from three clinical MRI protocols (gadolinium-enhanced T1-weighted, T2-weighted, and fluid-attenuated inversion recovery). The scans were retrospectively collected from the year 2000 until May 2019. Reference standard annotations of the tumor habitat, including enhancing tumor, edema, and cystic core plus nonenhancing tumor subcompartments, were performed by two experienced neuroradiologists. Preprocessing included registration to age-appropriate atlases, skull stripping, bias correction, and intensity matching. The two models were trained as follows: (a) the transfer learning nnU-Net model was pretrained on an adult glioma cohort (n = 484) and fine-tuned on medulloblastoma studies using Models Genesis and (b) the direct deep learning nnU-Net model was trained directly on the medulloblastoma datasets, across fivefold cross-validation. Model robustness was evaluated on the three datasets when using different combinations of training and test sets, with data from two sites at a time used for training and data from the third site used for testing. Results Analysis on the three test sites yielded Dice scores of 0.81, 0.86, and 0.86 and 0.80, 0.86, and 0.85 for tumor habitat; 0.68, 0.84, and 0.77 and 0.67, 0.83, and 0.76 for enhancing tumor; 0.56, 0.71, and 0.69 and 0.56, 0.71, and 0.70 for edema; and 0.32, 0.48, and 0.43 and 0.29, 0.44, and 0.41 for cystic core plus nonenhancing tumor for the transfer learning and direct nnU-Net models, respectively. The models were largely robust to site-specific variations. Conclusion nnU-Net segmentation models hold promise for accurate, robust automated delineation of medulloblastoma tumor subcompartments, potentially leading to more effective radiation therapy planning in pediatric medulloblastoma. Keywords: Pediatrics, MR Imaging, Segmentation, Transfer Learning, Medulloblastoma, nnU-Net, MRI Supplemental material is available for this article. © RSNA, 2024 See also the commentary by Rudie and Correia de Verdier in this issue.

Analyzing the Spatial Distribution of Immune Cells in Lung Adenocarcinoma.

(1) Background: This study investigates the tumor immune microenvironment, focusing on immune cell distribution in lung adenocarcinoma. (2) Methods: We evaluated fifty cases of lung adenocarcinoma, and suitable areas for further studies were annotated on the histological slides. Two tumor cores per case were obtained, one from the tumor's center and another from its periphery, and introduced into three paraffin receptor blocks for optimized processing efficiency. The 4-micrometer-thick tissue microarray sections were stained for H&E and for CD68, CD163, CD8, CD4, and PD-L1; (3) Results: Our investigation revealed significant correlations between PD-L1 expression in tumor cells and the presence of CD163+ macrophages, between CD4+ cells and CD8+, CD68+, and CD163+ cells, and also between CD8+ T cells and CD163+ cells. Additionally, while we observed some differences in cellular components and densities between the tumor center and periphery, these differences were not statistically significant. However, distinct correlations between PD-L1 and immune cells in these regions were identified, suggesting spatial heterogeneity in the immune landscape. (4) Conclusions: These results emphasize the intricate interactions between immune cells and tumor cells in lung adenocarcinoma. Understanding patient spatial immune profile could improve patient selection for immunotherapy, ensuring that those most likely to benefit are identified.

Automated brain tumor diagnostics: Empowering neuro-oncology with deep learning-based MRI image analysis.

Brain tumors, characterized by the uncontrolled growth of abnormal cells, pose a significant threat to human health. Early detection is crucial for successful treatment and improved patient outcomes. Magnetic Resonance Imaging (MRI) is the primary diagnostic tool for brain tumors, providing detailed visualizations of the brain's intricate structures. However, the complexity and variability of tumor shapes and locations often challenge physicians in achieving accurate tumor segmentation on MRI images. Precise tumor segmentation is essential for effective treatment planning and prognosis. To address this challenge, we propose a novel hybrid deep learning technique, Convolutional Neural Network and ResNeXt101 (ConvNet-ResNeXt101), for automated tumor segmentation and classification. Our approach commences with data acquisition from the BRATS 2020 dataset, a benchmark collection of MRI images with corresponding tumor segmentations. Next, we employ batch normalization to smooth and enhance the collected data, followed by feature extraction using the AlexNet model. This involves extracting features based on tumor shape, position, shape, and surface characteristics. To select the most informative features for effective segmentation, we utilize an advanced meta-heuristics algorithm called Advanced Whale Optimization (AWO). AWO mimics the hunting behavior of humpback whales to iteratively search for the optimal feature subset. With the selected features, we perform image segmentation using the ConvNet-ResNeXt101 model. This deep learning architecture combines the strengths of ConvNet and ResNeXt101, a type of ConvNet with aggregated residual connections. Finally, we apply the same ConvNet-ResNeXt101 model for tumor classification, categorizing the segmented tumor into distinct types. Our experiments demonstrate the superior performance of our proposed ConvNet-ResNeXt101 model compared to existing approaches, achieving an accuracy of 99.27% for the tumor core class with a minimum learning elapsed time of 0.53 s.

Assessment of Tumor Cell Invasion and Radiotherapy Response in Experimental Glioma by Magnetic Resonance Elastography.

Gliomas are highly invasive brain neoplasms. MRI is the most important tool to diagnose and monitor glioma but has shortcomings. In particular, the assessment of tumor cell invasion is insufficient. This is a clinical dilemma, as recurrence can arise from MRI-occult glioma cell invasion. Tumor cell invasion, tumor growth and radiotherapy alter the brain parenchymal microstructure and thus are assessable by diffusion tensor imaging (DTI) and MR elastography (MRE). Experimental, animal model. Twenty-three male NMRI nude mice orthotopically implanted with S24 patient-derived glioma cells (experimental mice) and 9 NMRI nude mice stereotactically injected with 1 μL PBS (sham-injected mice). 2D and 3D T2-weighted rapid acquisition with refocused echoes (RARE), 2D echo planar imaging (EPI) DTI, 2D multi-slice multi-echo (MSME) T2 relaxometry, 3D MSME MRE at 900 Hz acquired at 9.4 T (675 mT/m gradient strength). Longitudinal 4-weekly imaging was performed for up to 4 months. Tumor volume was assessed in experimental mice (n = 10 treatment-control, n = 13 radiotherapy). The radiotherapy subgroup and 5 sham-injected mice underwent irradiation (3 × 6 Gy) 9 weeks post-implantation/sham injection. MRI-/MRE-parameters were assessed in the corpus callosum and tumor core/injection tract. Imaging data were correlated to light sheet microscopy (LSM) and histology. Paired and unpaired t-tests, a P-value ≤0.05 was considered significant. From week 4 to 8, a significant callosal stiffening (4.44 ± 0.22 vs. 5.31 ± 0.29 kPa) was detected correlating with LSM-proven tumor cell invasion. This was occult to all other imaging metrics. Histologically proven tissue destruction in the tumor core led to an increased T2 relaxation time (41.65 ± 0.34 vs. 44.83 ± 0.66 msec) and ADC (610.2 ± 12.27 vs. 711.2 ± 13.42 × 10-6 mm2/s) and a softening (5.51 ± 0.30 vs. 4.24 ± 0.29 kPa) from week 8 to 12. Radiotherapy slowed tumor progression. MRE is promising for the assessment of key glioma characteristics. NA TECHNICAL EFFICACY: Stage 2.

T cell factor 1 (TCF-1) defines T cell differentiation in colorectal cancer

The presence of precursor to exhausted (Tpex) CD8+ Tcells is important to maintain robust immunity following treatment with immune checkpoint inhibition (ICI). Impressive responses to ICI are emerging in patients with stage II-III mismatch repair (MMR)-deficient (dMMR) colorectal cancer (CRC). We found 64% of dMMR and 15% of mismatch repair-proficient (pMMR) stage III CRCs had a high frequency of tumor infiltrating lymphocytes (TIL-hi). Furthermore, expression of TCF-1 (Tcf7) by CD8+ Tcells predicted improved patient prognosis and Tpex cells (CD3+CD8+TCF-1+PD-1+) were abundant within lymphoid aggregatesof stage III CRCs. In contrast, CD3+CD8+TCF-1-PD-1+ cells were more abundant at the invasive front and tumor core, while γδ Tcells were equally abundant in all tumor areas. Interestingly, no differencesin the frequency of Tpex cells were observed between TIL-hi dMMR and TIL-hi pMMR CRCs. Therefore, Tpex cell function and ICI response rates in TIL-hi CRC warrants further investigation.

Changes in perfusion and permeability in glioblastoma model induced by the anti-angiogenic agents cediranib and thalidomide.

The poor delivery of drugs to infiltrating tumor cells contributes to therapeutic failure in glioblastoma. During the early phase of an anti-angiogenic treatment, a remodeling of the tumor vasculature could occur, leading to a more functional vessel network that could enhance drug delivery. However, the restructuration of blood vessels could increase the proportion of normal endothelial cells that could be a barrier for the free diffusion of drugs. The net balance, in favor or not, of a better delivery of compounds during the course of an antiangiogenic treatment remains to be established. This study explored whether cediranib and thalidomide could modulate perfusion and vessel permeability in the brain U87 tumor mouse model. The dynamic evolution of the diffusion of agents outside the tumor core using the fluorescent dye Evans Blue in histology and Gd-DOTA using dynamic contrast-enhanced (DCE)-MRI. CD31 labelling of endothelial cells was used to measure the vascular density. Cediranib and thalidomide effectively reduced tumor size over time. The accessibility of Evans Blue outside the tumor core continuously decreased over time. The vascular density was significantly decreased after treatment while the proportion of normal vessels remained unchanged over time. In contrast to histological studies, DCE-MRI did not tackle any significant change in hemodynamic parameters, in the core or margins of the tumor, whatever the parameter used or the pharmacokinetic model used. While cediranib and thalidomide were effective in decreasing the tumor size, they were ineffective in transiently increasing the delivery of agents in the core and the margins of the tumor.

Stereotactic centralized ablative radiation therapy: A novel methodology in treating bulky tumor and its technical realization.

159 Background: Bulky tumors pose significant challenges to traditional treatment modalities such as surgery, chemotherapy, and conventional radiotherapy. This study introduces a novel therapeutic approach named as Stereotactic-Core-Ablative-Radiation-Therapy (SCART), designed to deliver an ablative dose to a substantial core of the bulky tumor while quickly decreasing to a lower dose at the tumor periphery. Methods: A SCART-Treatment-Volume (STV) at the core of the GTV was predefined to administer an ablative radiation dose, aiming to induce DNA damage in cancer cells and potentially trigger biological cancer-killing effects. Utilizing Linac-based VMAT or Cyberknife techniques with 6MV photon beams, multiple radiation fields intersect and optimize at STV, generating an ablative dose at the STV. The dose rapidly diminishes to a safe level at the edge of GTV sparing the surrounding tissue. In the phase-1 trial, nineteen patients with 21 bulky tumors enrolled, receiving SCART at five dose levels (15GyX1, 15GyX2, 15GyX3, 18GyX3, 21GyX3, and 24GyX3), while maintaining the GTV's peripheral dose at 5Gy each fraction. Results: All patients completed treatment with average beam-on time of 8.9min and average treatment time of 18.5min. Mean follow-up time is 15.4 month. No grade-III or higher toxicity was observed. 7/19 patients still survive, with the overall survival of 40% at 30 months. Mean tumor volume shrinks by 60% between initial 301cc and post-SCART volumes of 118cc. Long-term follow-up revealed that 14/21 tumors achieved Partial Response, 2/21 Complete-Response, 3/21 Stable-Disease, and 1/21 Progressive-Disease, leading to an encouraging local control of 95%. Conclusions: SCART emerges as a safe and effective strategy for treating bulky malignant tumors, demonstrating excellent local control and overall survival. Multiple treatment courses were feasible. The results from phase-1 study suggest that SCART could revolutionize the treatment landscape for bulky tumors, offering a promising avenue for further exploration and application in clinical practice.

Microtubule-associated NAV3 regulates invasive phenotypes in glioblastoma cells.

Glioblastomas are aggressive brain tumors for which effective therapy is still lacking, resulting in dismal survival rates. These tumors display significant phenotypic plasticity, harboring diverse cell populations ranging from tumor core cells to dispersed, highly invasive cells. Neuron navigator 3 (NAV3), a microtubule-associated protein affecting microtubule growth and dynamics, is downregulated in various cancers, including glioblastoma, and has thus been considered a tumor suppressor. In this study, we challenge this designation and unveil distinct expression patterns of NAV3 across different invasion phenotypes. Using glioblastoma cell lines and patient-derived glioma stem-like cell cultures, we disclose an upregulation of NAV3 in invading glioblastoma cells, contrasting with its lower expression in cells residing in tumor spheroid cores. Furthermore, we establish an association between low and high NAV3 expression and the amoeboid and mesenchymal invasive phenotype, respectively, and demonstrate that overexpression of NAV3 directly stimulates glioblastoma invasive behavior in both 2D and 3D environments. Consistently, we observed increased NAV3 expression in cells migrating along blood vessels in mouse xenografts. Overall, our results shed light on the role of NAV3 in glioblastoma invasion, providing insights into this lethal aspect of glioblastoma behavior.

Modality redundancy for MRI-based glioblastoma segmentation

PurposeAutomated glioblastoma segmentation from magnetic resonance imaging is generally performed on a four-modality input, including T1, contrast T1, T2 and FLAIR. We hypothesize that information redundancy is present within these image combinations, which can possibly reduce a model’s performance. Moreover, for clinical applications, the risk of encountering missing data rises as the number of required input modalities increases. Therefore, this study aimed to explore the relevance and influence of the different modalities used for MRI-based glioblastoma segmentation.MethodsAfter the training of multiple segmentation models based on nnU-Net and SwinUNETR architectures, differing only in their amount and combinations of input modalities, each model was evaluated with regard to segmentation accuracy and epistemic uncertainty.ResultsResults show that T1CE-based segmentation (for enhanced tumor and tumor core) and T1CE-FLAIR-based segmentation (for whole tumor and overall segmentation) can reach segmentation accuracies comparable to the full-input version. Notably, the highest segmentation accuracy for nnU-Net was found for a three-input configuration of T1CE-FLAIR-T1, suggesting the confounding effect of redundant input modalities. The SwinUNETR architecture appears to suffer less from this, where said three-input and the full-input model yielded statistically equal results.ConclusionThe T1CE-FLAIR-based model can therefore be considered as a minimal-input alternative to the full-input configuration. Addition of modalities beyond this does not statistically improve and can even deteriorate accuracy, but does lower the segmentation uncertainty.

Deep learning models for rapid discrimination of high-grade gliomas from solitary brain metastases using multi-plane T1-weighted contrast-enhanced (T1CE) images.

High-grade gliomas (HGG) and solitary brain metastases (SBM) are two common types of brain tumors in middle-aged and elderly patients. HGG and SBM display a high degree of similarity on magnetic resonance imaging (MRI) images. Consequently, differential diagnosis using preoperative MRI remains challenging. This study developed deep learning models that used pre-operative T1-weighted contrast-enhanced (T1CE) MRI images to differentiate between HGG and SBM before surgery. By comparing various convolutional neural network models using T1CE image data from The First Medical Center of the Chinese PLA General Hospital and The Second People's Hospital of Yibin (Data collection for this study spanned from January 2016 to December 2023), it was confirmed that the GoogLeNet model exhibited the highest discriminative performance. Additionally, we evaluated the individual impact of the tumoral core and peritumoral edema regions on the network's predictive performance. Finally, we adopted a slice-based voting method to assess the accuracy of the validation dataset and evaluated patient prediction performance on an additional test dataset. The GoogLeNet model, in a five-fold cross-validation using multi-plane T1CE slices (axial, coronal, and sagittal) from 180 patients, achieved an average patient accuracy of 92.78%, a sensitivity of 95.56%, and a specificity of 90.00%. Moreover, on an external test set of 29 patients, the model achieved an accuracy of 89.66%, a sensitivity of 90.91%, and a specificity of 83.33%, with an area under the curve of 0.939 [95% confidence interval (CI): 0.842-1.000]. GoogLeNet performed better than previous methods at differentiating HGG from SBM, even for core and peritumoral edema in both. HGG and SBM could be fast screened using this end-to-end approach, improving workflow for both tumor treatments.

Evaluation of CD3 and CD8 T-Cell Immunohistochemistry for Prognostication and Prediction of Benefit From Adjuvant Chemotherapy in Early-Stage Colorectal Cancer Within the QUASAR Trial.

High densities of tumor infiltrating CD3 and CD8 T-cells are associated with superior prognosis in colorectal cancer (CRC). Their value as predictors of benefit from adjuvant chemotherapy is uncertain. Tumor tissue from 868 patients in the QUASAR trial (adjuvant fluorouracil/folinic acid v observation in stage II/III CRC) was analyzed by CD3 and CD8 immunohistochemistry. Pathologists, assisted by artificial intelligence, calculated CD3 and CD8 cell densities (cells/mm2) in the core tumor (CT) and invasive margin (IM). Participants were randomly partitioned into training and validation sets. The primary outcome was recurrence-free interval (RFI), 2-year RFI for assessment of biomarker-treatment interactions. Maximum-likelihood methods identified optimal high-risk/low-risk group cutpoints in the training set. Prognostic analyses were repeated in the validation set. In the training set, the recurrence rate in the high-risk group was twice that in the low-risk group for all measures (CD3-CT: rate ratio [RR], 2.00, P = .0008; CD3-IM: 2.38, P < .00001; CD8-CT: 2.17, P = .0001; CD8-IM: 2.13, P = .0001). This was closely replicated in the validation set (RR, 1.96, 1.79, 1.72, 1.72, respectively). In multivariate analyses, prognostic effects were similar in colon and rectal cancers, and in stage II and III disease. Proportional reductions in recurrence with adjuvant chemotherapy were of similar magnitude in the high- and low-recurrence risk groups. Combining information from CD3-IM and CD3-CT (CD3 Score) generated high-, intermediate-, and low-risk groups with numbers needed to treat (NNTs) to prevent one disease recurrence being 11, 21, and 36, respectively. Recurrence rates in the high-risk CD3/CD8 groups are twice those in the low-risk groups. Proportional reductions with chemotherapy are similar, allowing NNTs derived in QUASAR to be updated using contemporary, nonrandomized data sets.

Estimation of the injection criteria for magnetic hyperthermia therapy based on tumor morphology

Intratumoral multi-injection strategy enhances the efficacy of magnetic nanoparticle hyperthermia therapy (MNPH). In this study, criteria for the selection of injections and their location depending on the tumor shape/geometry are developed. The developed strategy is based on the thermal dosimetry results of different invasive 3D tumor models during MNPH simulation. MNPH simulations are conducted on physical tumor tissue models encased within healthy tissue. The tumor shapes are geometrically divided into a central tumor region containing maximum tumor volume and a peripheral tumor portion protruding in any random direction. The concepts of core and invasive radius are used to geometrically divide the tumor volume. Primary & secondary injections are used to inject MNP fluid into these respective tumor regions based on the invasiveness of the tumor. The optimization strategy is devised based on the zone of influence of primary & secondary injection. Results indicate that the zone of influence of secondary injection lies between 0.7 and 0.8 times the radial distance between the center of the tumor core and branch node point (extreme far endpoint on the invasive tumor surface). Additionally, the multi-injection strategy is more effective when the protrusion volume exceeds 10% of the total volume. The proposed algorithm is used to devise multi-injection strategies for arbitrarily shaped tumors and will assist in pre-planning magnetic nanoparticle hyperthermia therapy.

Multiscale segmentation net for segregating heterogeneous brain tumors: Gliomas on multimodal MR images

In this research, the 3D volumetric segmentation of heterogeneous brain tumors such as Gliomas- anaplastic astrocytoma, and Glioblastoma Multiforme (GBM) is performed to extract enhancing tumor (ET), whole tumor (WT), and tumor core (TC) regions using T1, T2, and FLAIR images. Therefore, a deep learning-based encoder-decoder architecture named “MS-SegNet” using 3D multi-scale convolutional layers is proposed. The proposed architecture employs multi-scale feature extraction (MS-FE) block the filter size 3 × 3 × 3 to extract confined information like tumor boundary and edges of necrotic part. The filter of size 5 × 5 × 5 focuses on varied features like shape, size, and location of tumor region with edema. The local and global features from different MR modalities are extracted for segmenting thin and meshed boundaries of heterogeneous tumors between anatomical sub-regions like peritumoral edema, enhancing tumor, and necrotic tumor core. The learning parameters on introducing the MS-FE block are reduced to 10 million, which is much less than other architectures like 3D-Unet which takes into consideration 27 million features leading to the consumption of less computational power. A customized loss function is also prophesied based on a combination of dice loss and focal loss along with metrics such as accuracy and Intersection over Union (IoU) i.e. the overlapping of ground truth mask and predicted value for addressing the class imbalance problem. For evaluating the efficacy of the proposed method, four evaluation metrics such as Dice Coefficient (DSC), Sensitivity, Specificity, and Hausdorff95 distance (H95) are employed for analyzing the model's overall performance. It is observed that the proposed MS-SegNet architecture achieved the DSC of 0.81, 0.91, and 0.83 on BraTS 2020; 0.86, 0.92, and 0.84 on BraTS 2021 for ET, WT, and TC respectively. The developed model is also tested on a real-time dataset collected from the Post Graduate Institute of Medical Education & Research (PGIMER), Chandigarh. The DSC of 0.79, 0.76, and 0.68 for ET, WT, and TC respectively on the real-time dataset. These findings show that deep learning models with enhanced feature extraction capabilities can be readily trained to attain high accuracy in segmenting heterogeneous brain tumors and hold promising results. In the future, other tumor datasets will be explored for the detection and treatment planning of brain tumors to check the effectiveness of the model in real-world healthcare environments.

SwinVNETR: Swin V-net Transformer with non-local block for volumetric MRI Brain Tumor Segmentation

Brain Tumor Segmentation (BTS) and classification are important and growing research fields. Magnetic resonance imaging (MRI) is commonly used in the diagnosis of brain tumours owing to its low radiation exposure and high image quality. One of the current subjects in the field of medical imaging is how to quickly and precisely segment MRI scans of brain tumours. Unfortunately, most existing brain tumour segmentation algorithms use inadequate 2D picture segmentation methods and fail to capture the spatial correlation between features. In this study, we propose a segmentation model (SwinVNETR) Swin V-NetTRansformer-based architecture with a non-local block. This model was trained using the Brain Tumor Segmentation Challenge BraTS 2021 dataset. The Dice similarity coefficients for the enhanced tumour (ET), whole tumour (WT), and tumour core (TC) are 0.84, 0.91, and 0.87, respectively. By leveraging this methodology, we can segment brain tumours more accurately than ever before. In conclusion, we present the findings of our model through the application of the Grad-CAM methodology, an eXplainable Artificial Intelligence (XAI) technique utilized to elucidate the insights derived from the model, which helped in better understanding; doctors can better diagnose and treat patients with brain tumours.

CMAF-Net: a cross-modal attention fusion-based deep neural network for incomplete multi-modal brain tumor segmentation.

The information between multimodal magnetic resonance imaging (MRI) is complementary. Combining multiple modalities for brain tumor image segmentation can improve segmentation accuracy, which has great significance for disease diagnosis and treatment. However, different degrees of missing modality data often occur in clinical practice, which may lead to serious performance degradation or even failure of brain tumor segmentation methods relying on full-modality sequences to complete the segmentation task. To solve the above problems, this study aimed to design a new deep learning network for incomplete multimodal brain tumor segmentation. We propose a novel cross-modal attention fusion-based deep neural network (CMAF-Net) for incomplete multimodal brain tumor segmentation, which is based on a three-dimensional (3D) U-Net architecture with encoding and decoding structure, a 3D Swin block, and a cross-modal attention fusion (CMAF) block. A convolutional encoder is initially used to extract the specific features from different modalities, and an effective 3D Swin block is constructed to model the long-range dependencies to obtain richer information for brain tumor segmentation. Then, a cross-attention based CMAF module is proposed that can deal with different missing modality situations by fusing features between different modalities to learn the shared representations of the tumor regions. Finally, the fused latent representation is decoded to obtain the final segmentation result. Additionally, channel attention module (CAM) and spatial attention module (SAM) are incorporated into the network to further improve the robustness of the model; the CAM to help focus on important feature channels, and the SAM to learn the importance of different spatial regions. Evaluation experiments on the widely-used BraTS 2018 and BraTS 2020 datasets demonstrated the effectiveness of the proposed CMAF-Net which achieved average Dice scores of 87.9%, 81.8%, and 64.3%, as well as Hausdorff distances of 4.21, 5.35, and 4.02 for whole tumor, tumor core, and enhancing tumor on the BraTS 2020 dataset, respectively, outperforming several state-of-the-art segmentation methods in missing modalities situations. The experimental results show that the proposed CMAF-Net can achieve accurate brain tumor segmentation in the case of missing modalities with promising application potential.

Prognostic relevance of high expression of kynurenine pathway markers in glioblastoma

Glioblastoma (GBM) continues to exhibit a discouraging survival rate despite extensive research into new treatments. One factor contributing to its poor prognosis is the tumor's immunosuppressive microenvironment, in which the kynurenine pathway (KP) plays a significant role. This study aimed to explore how KP impacts the survival of newly diagnosed GBM patients. We examined tissue samples from 108 GBM patients to assess the expression levels of key KP markers—tryptophan 2,3-dioxygenase (TDO2), indoleamine 2,3-dioxygenase (IDO1/2), and the aryl hydrocarbon receptor (AhR). Using immunohistochemistry and QuPath software, three tumor cores were analyzed per patient to evaluate KP marker expression. Kaplan–Meier survival analysis and stepwise multivariate Cox regression were used to determine the effect of these markers on patient survival. Results showed that patients with high expression of TDO2, IDO1/2, and AhR had significantly shorter survival times. This finding held true even when controlling for other known prognostic variables, with a hazard ratio of 3.393 for IDO1, 2.775 for IDO2, 1.891 for TDO2, and 1.902 for AhR. We suggest that KP markers could serve as useful tools for patient stratification, potentially guiding future immunomodulating trials and personalized treatment approaches for GBM patients.

The prognostic effect of mechanical, ultrastructural, and ECM signatures in glioblastoma core and rim.

Glioblastoma (GBM) is a highly invasive, aggressive brain cancer that carries a median survival of 15 months and is resistant to standard therapeutics. Recent studies have demonstrated that intratumoral heterogeneity plays a critical role in promoting resistance by mediating tumor adaptation through microenvironmental cues. GBM can be separated into two distinct regions-a core and a rim, which are thought to drive specific aspects of tumor evolution. These differences in tumor progression are regulated by the diverse biomolecular and biophysical signals in these regions, but the acellular biophysical characteristics remain poorly described. This study investigates the mechanical and ultrastructural characteristics of the tumor extracellular matrix (ECM) in patient-matched GBM core and rim tissues. Seven patient-matched tumor core and rim samples and one non-neoplastic control were analyzed using atomic force microscopy, scanning electron microscopy, and immunofluorescence imaging to quantify mechanical, ultrastructural, and ECM composition changes. The results reveal significant differences in biophysical parameters between GBM core, rim, and non-neoplastic tissues. The GBM core is stiffer, denser, and is rich in ECM proteins hyaluronic acid and tenascin-C when compared to tumor rim and non-neoplastic tissues. These alterations are intimately related and have prognostic effect with stiff, dense tissue correlating with longer progression-free survival. These findings reveal new insights into the spatial heterogeneity of biophysical parameters in the GBM tumor microenvironment and identify a set of characteristics that may correlate with patient prognosis. In the long term, these characteristics may aid in the development of strategies to combat therapeutic resistance.

Quantifying spatial CXCL9 distribution with image analysis predicts improved prognosis of triple-negative breast cancer.

Background: The C-X-C motif chemokine ligand 9 (CXCL9) plays a pivotal role in tumor immunity by recruiting and activating immune cells. However, the relationship between CXCL9 expression and prognosis in triple-negative breast cancer (TNBC) is unclear. Methods: We investigated CXCL9 mRNA expression, clinicopathological features, and prognosis in TNBC patients. We also used computational image analysis to quantify and assess the distribution of CXCL9 protein in the tumor core (TC) and invasive margin (IM). Results: CXCL9 mRNA expression was significantly higher in TNBC tumors compared to normal tissue (p < 0.001) and was associated with smaller tumors (p = 0.022) and earlier stages (p = 0.033). High CXCL9 mRNA expression was correlated with improved overall survival (OS) in three independent cohorts (all p < 0.05). In a separate analysis, low CXCL9 protein expression was associated with increased lymph node metastasis (p = 0.018 and p = 0.036). High CXCL9 protein expression in the TC, IM, or both was associated with prolonged OS (all p < 0.001). Conclusion: High CXCL9 expression, at both the mRNA and protein levels, is associated with improved prognosis in TNBC patients. CXCL9 expression in the TC and/or IM may be an independent prognostic factor.

IMG-28. AUTOMATIC BRAIN TUMOR VOLUMETRIC ANALYSIS IN MAGNETIC RESONANCE IMAGING GENERALIZABLE TO PEDIATRIC NEURO-ONCOLOGY

Abstract BACKGROUND The prognosis of brain tumors is variable in clinical practice if it only relies on human interpretation of magnetic resonance imaging (MRI). The automatic segmentation of brain tumors in MRI enables quantitative analysis in support of clinical trials and personalized patient care. We developed benchmarked deep learning-based tools that are generalizable to the volumetric quantification of various tumor types across diverse populations. METHODS We participated in the well-established international brain tumor segmentation challenge (BraTS 2023) and benchmarking competition. The challenge made available 4,500 multi-national brain tumor cases with multi-sequence MRIs, including pediatric high-grade gliomas (PED), i.e., high-grade astrocytoma and diffuse midline glioma, and adult gliomas, brain metastases (MET) and intracranial meningiomas (MEN). Each case comprises four MRI volumes: T1, contrast-enhanced T1, T2, and T2-FLAIR. Manual segmentations were provided to establish ground truth for enhancing tumor (ET), tumor core (TC), and whole tumor (WT). Our framework used a model ensemble strategy based on two state-of-the-art deep learning models: a convolutional neural network (nnU-Net) and a vision transformer (Swin UNETR) and was tested for broader applicability across multiple tumor types. The framework was trained on 99, 1,000, and 165 cases and validated on 45, 141, and 31 unseen cases for PED, MEN, and MET, respectively. Automatic segmentations were evaluated by lesion-wise volume overlap (Dice similarity score, DSC) and Hausdorff distance (HD). RESULTS In the evaluation on independent unseen test datasets, our automatic tool was ranked first for PED, third for MEN, and fourth for MET volumetric analysis. Our method resulted in PED lesion-wise DSC of 0.733, 0.782, 0.817 and HD (mm) of 75.93, 25.54, 24.18 for ET, TC, and WT, respectively. CONCLUSIONS These brain tumor volumetric analysis tools are readily available to be efficiently tested on diverse datasets. Automatic MRI analysis provides consistent quantitative data for multi-institutional protocols and clinical trials.