A novel semi-supervised methodology for extracting tumor type-specific MRS sources in human brain data

Tracking Therapy Response in Glioblastoma Using 1D Convolutional Neural Networks

BACKGROUND

Glioblastoma (GB) is a malignant brain tumour that is challenging to treat, often relapsing even after aggressive therapy. Evaluating therapy response relies on magnetic resonance imaging (MRI) following the Response Assessment in Neuro-Oncology (RANO) criteria. However, early assessment is hindered by phenomena such as pseudoprogression and pseudoresponse. Magnetic resonance spectroscopy (MRS/MRSI) provides metabolomics information but is underutilised due to a lack of familiarity and standardisation.

METHODS

This study explores the potential of spectroscopic imaging (MRSI) in combination with several machine learning approaches, including one-dimensional convolutional neural networks (1D-CNNs), to improve therapy response assessment. Preclinical GB (GL261-bearing mice) were studied for method optimisation and validation.

RESULTS

The proposed 1D-CNN models successfully identify different regions of tumours sampled by MRSI, i.e., normal brain (N), control/unresponsive tumour (T), and tumour responding to treatment (R). Class activation maps using Grad-CAM enabled the study of the key areas relevant to the models, providing model explainability. The generated colour-coded maps showing the N, T and R regions were highly accurate (according to Dice scores) when compared against ground truth and outperformed our previous method.

CONCLUSIONS

The proposed methodology may provide new and better opportunities for therapy response assessment, potentially providing earlier hints of tumour relapsing stages.

Breast cancer patient characterisation and visualisation using deep learning and fisher information networks

Breast cancer is the most commonly diagnosed female malignancy globally, with better survival rates if diagnosed early. Mammography is the gold standard in screening programmes for breast cancer, but despite technological advances, high error rates are still reported. Machine learning techniques, and in particular deep learning (DL), have been successfully used for breast cancer detection and classification. However, the added complexity that makes DL models so successful reduces their ability to explain which features are relevant to the model, or whether the model is biased. The main aim of this study is to propose a novel visualisation to help characterise breast cancer patients using Fisher Information Networks on features extracted from mammograms using a DL model. In the proposed visualisation, patients are mapped out according to their similarities and can be used to study new patients as a ‘patient-like-me’ approach. When applied to the CBIS-DDSM dataset, it was shown that it is a competitive methodology that can (i) facilitate the analysis and decision-making process in breast cancer diagnosis with the assistance of the FIN visualisations and ‘patient-like-me’ analysis, and (ii) help improve diagnostic accuracy and reduce overdiagnosis by identifying the most likely diagnosis based on clinical similarities with neighbouring patients.

Extraction of artefactual MRS patterns from a large database using non-negative matrix factorization

Despite the success of automated pattern recognition methods in problems of human brain tumor diagnostic classification, limited attention has been paid to the issue of automated data quality assessment in the field of MRS for neuro-oncology. Beyond some early attempts to address this issue, the current standard in practice is MRS quality control through human (expert-based) assessment. One aspect of automatic quality control is the problem of detecting artefacts in MRS data. Artefacts, whose variety has already been reviewed in some detail and some of which may even escape human quality control, have a negative influence in pattern recognition methods attempting to assist tumor characterization. The automatic detection of MRS artefacts should be beneficial for radiology as it guarantees more reliable tumor characterizations, as well as the development of more robust pattern recognition-based tumor classifiers and more trustable MRS data processing and analysis pipelines. Feature extraction methods have previously been used to help distinguishing between good and bad quality spectra to apply subsequent supervised pattern recognition techniques. In this study, we apply feature extraction differently and use a variant of a method for blind source separation, namely Convex Non-Negative Matrix Factorization, to unveil MRS signal sources in a completely unsupervised way. We hypothesize that, while most sources will correspond to the different tumor patterns, some of them will reflect signal artefacts. The experimental work reported in this paper, analyzing a combined short and long echo time ¹ H-MRS database of more than 2000 spectra acquired at 1.5T and corresponding to different tumor types and other anomalous masses, provides a first proof of concept that points to the possible validity of this approach.

Anti-PD-1 Immunotherapy in Preclinical GL261 Glioblastoma: Influence of Therapeutic Parameters and Non-Invasive Response Biomarker Assessment with MRSI-Based Approaches

Glioblastomas (GBs) are malignant brain tumours with poor prognosis even after aggressive therapy. Programmed cell death-1 (PD-1) immune checkpoint blockade is a promising strategy in many types of cancer, but its therapeutic effects in GB remain low and associated with immune infiltration. Previous work suggests that oscillations of magnetic resonance spectroscopic imaging (MRSI)-based response pattern with chemotherapy could act as a biomarker of efficient immune system attack onto GBs. The presence of such oscillations with other monotherapies such as anti-PD-1 would reinforce its monitoring potential. Here, we confirm that the oscillatory behaviour of the response biomarker is also detected in mice treated with anti PD-1 immunotherapy both in combination with temozolomide and as monotherapy. This indicates that the spectral pattern changes observed during therapy response are shared by different therapeutic strategies, provided the host immune system is elicited and able to productively attack tumour cells. Moreover, the participation of the immune system in response is also supported by the rate of cured animals observed with different therapeutic strategies (in the range of 50–100% depending on the treatment), which also held long-term immune memory against tumour cells re-challenge. Taken together, our findings open the way for a translational use of the MRSI-based biomarker in patient-tailored GB therapy, including immunotherapy, for which reliable non-invasive biomarkers are still missing.

Anti-tumour immune response in GL261 glioblastoma generated by Temozolomide Immune-Enhancing Metronomic Schedule monitored with MRSI-based nosological images

Glioblastomas (GB) are brain tumours with poor prognosis even after aggressive therapy. Improvements in both therapeutic and follow-up strategies are urgently needed. In previous work we described an oscillatory pattern of response to Temozolomide (TMZ) using a standard administration protocol, detected through MRSI-based machine learning approaches. In the present work, we have introduced the Immune-Enhancing Metronomic Schedule (IMS) with an every 6-d TMZ administration at 60 mg/kg and investigated the consistence of such oscillatory behaviour. A total of n = 17 GL261 GB tumour-bearing C57BL/6j mice were studied with MRI/MRSI every 2 d, and the oscillatory behaviour (6.2 ± 1.5 d period from the TMZ administration day) was confirmed during response. Furthermore, IMS-TMZ produced significant improvement in mice survival (22.5 ± 3.0 d for controls vs 135.8 ± 78.2 for TMZ-treated), outperforming standard TMZ treatment. Histopathological correlation was investigated in selected tumour samples (n = 6) analyzing control and responding fields. Significant differences were found for CD3+ cells (lymphocytes, 3.3 ± 2.5 vs 4.8 ± 2.9, respectively) and Iba-1 immunostained area (microglia/macrophages, 16.8% ± 9.7% and 21.9% ± 11.4%, respectively). Unexpectedly, during IMS-TMZ treatment, tumours from some mice (n = 6) fully regressed and remained undetectable without further treatment for 1 mo. These animals were considered "cured" and a GL261 re-challenge experiment performed, with no tumour reappearance in five out of six cases. Heterogeneous therapy response outcomes were detected in tumour-bearing mice, and a selected group was investigated (n = 3 non-responders, n = 6 relapsing tumours, n = 3 controls). PD-L1 content was found ca. 3-fold increased in the relapsing group when comparing with control and non-responding groups, suggesting that increased lymphocyte inhibition could be associated to IMS-TMZ failure. Overall, data suggest that host immune response has a relevant role in therapy response/escape in GL261 tumours under IMS-TMZ therapy. This is associated to changes in the metabolomics pattern, oscillating every 6 d, in agreement with immune cycle length, which is being sampled by MRSI-derived nosological images.

Embedding MRI information into MRSI data source extraction improves brain tumour delineation in animal models

Glioblastoma is the most frequent malignant intra-cranial tumour. Magnetic resonance imaging is the modality of choice in diagnosis, aggressiveness assessment, and follow-up. However, there are examples where it lacks diagnostic accuracy. Magnetic resonance spectroscopy enables the identification of molecules present in the tissue, providing a precise metabolomic signature. Previous research shows that combining imaging and spectroscopy information results in more accurate outcomes and superior diagnostic value. This study proposes a method to combine them, which builds upon a previous methodology whose main objective is to guide the extraction of sources. To this aim, prior knowledge about class-specific information is integrated into the methodology by setting the metric of a latent variable space where Non-negative Matrix Factorisation is performed. The former methodology, which only used spectroscopy and involved combining spectra from different subjects, was adapted to use selected areas of interest that arise from segmenting the T2-weighted image. Results showed that embedding imaging information into the source extraction (the proposed semi-supervised analysis) improved the quality of the tumour delineation, as compared to those obtained without this information (unsupervised analysis). Both approaches were applied to pre-clinical data, involving thirteen brain tumour-bearing mice, and tested against histopathological data. On results of twenty-eight images, the proposed Semi-Supervised Source Extraction (SSSE) method greatly outperformed the unsupervised one, as well as an alternative semi-supervised approach from the literature, with differences being statistically significant. SSSE has proven successful in the delineation of the tumour, while bringing benefits such as 1) not constricting the metabolomic-based prediction to the image-segmented area, 2) ability to deal with signal-to-noise issues, 3) opportunity to answer specific questions by allowing researchers/radiologists define areas of interest that guide the source extraction, 4) creation of an intra-subject model and avoiding contamination from inter-subject overlaps, and 5) extraction of meaningful, good-quality sources that adds interpretability, conferring validation and better understanding of each case.

ACRIN 6684: Multicenter, phase II assessment of tumor hypoxia in newly diagnosed glioblastoma using magnetic resonance spectroscopy

A multi-center imaging trial by the American College of Radiology Imaging Network (ACRIN) "A Multicenter, phase II assessment of tumor hypoxia in glioblastoma using 18F Fluoromisonidazole (FMISO) with PET and MRI (ACRIN 6684)", was conducted to assess hypoxia in patients with glioblastoma (GBM). The aims of this study were to support the role of proton magnetic resonance spectroscopic imaging (1H MRSI) as a prognostic marker for brain tumor patients in multi-center clinical trials. Seventeen participants from four sites had analyzable 3D MRSI datasets acquired on Philips, GE or Siemens scanners at either 1.5T or 3T. MRSI data were analyzed using LCModel to quantify metabolites N-acetylaspartate (NAA), creatine (Cr), choline (Cho), and lactate (Lac). Receiver operating characteristic curves for NAA/Cho, Cho/Cr, lactate/Cr, and lactate/NAA were constructed for overall survival at 1-year (OS-1) and 6-month progression free survival (PFS-6). The OS-1 for the 17 evaluable patients was 59% (10/17). Receiver operating characteristic analyses found the NAA/Cho in tumor (AUC = 0.83, 95% CI: 0.61 to 1.00) and in peritumoral regions (AUC = 0.95, 95% CI 0.85 to 1.00) were predictive for survival at 1 year. PFS-6 was 65% (11/17). Neither NAA/Cho nor Cho/Cr was effective in predicting 6-month progression free survival. Lac/Cr in tumor was a significant negative predictor of PFS-6, indicating that higher lactate/Cr levels are associated with poorer outcome. (AUC = 0.79, 95% CI: 0.54 to 1.00). In conclusion, despite the small sample size in the setting of a multi-center trial comprising different vendors, field strengths, and varying levels of expertise at data acquisition, MRS markers NAA/Cho, Lac/Cr and Lac/NAA predicted overall survival at 1 year and 6-month progression free survival. This study validates that MRSI may be useful in evaluating the prognosis in glioblastoma and should be considered for incorporating into multi-center clinical trials.

Metronomic treatment in immunocompetent preclinical GL261 glioblastoma: effects of cyclophosphamide and temozolomide

Glioblastoma (GBM) causes poor survival in patients even when applying aggressive treatment. Temozolomide (TMZ) is the standard chemotherapeutic choice for GBM treatment, but resistance always ensues. In previous years, efforts have focused on new therapeutic regimens with conventional drugs to activate immune responses that may enhance tumor regression and prevent regrowth, for example the "metronomic" approaches. In metronomic scheduling studies, cyclophosphamide (CPA) in GL261 GBM growing subcutaneously in C57BL/6 mice was shown not only to activate antitumor CD8⁺ T-cell response, but also to induce long-term specific T-cell tumor memory. Accordingly, we have evaluated whether metronomic CPA or TMZ administration could increase survival in orthotopic GL261 in C57BL/6 mice, an immunocompetent model. Longitudinal in vivo studies with CPA (140 mg/kg) or TMZ (range 140-240 mg/kg) metronomic administration (every 6 days) were performed in tumor-bearing mice. Tumor evolution was monitored at 7 T with MRI (T₂ -weighted, diffusion-weighted imaging) and MRSI-based nosological images of response to therapy. Obtained results demonstrated that both treatments resulted in increased survival (38.6 ± 21.0 days, n = 30) compared with control (19.4 ± 2.4 days, n = 18). Best results were obtained with 140 mg/kg TMZ (treated, 44.9 ± 29.0 days, n = 12, versus control, 19.3 ± 2.3 days, n = 12), achieving a longer survival rate than previous group work using three cycles of TMZ therapy at 60 mg/kg (33.9 ± 11.7 days, n = 38). Additional interesting findings were, first, clear edema appearance during chemotherapeutic treatment, second, the ability to apply the semi-supervised source analysis previously developed in our group for non-invasive TMZ therapy response monitoring to detect CPA-induced response, and third, the necropsy findings in mice cured from GBM after high TMZ cumulative dosage (980-1400 mg/kg), which demonstrated lymphoma incidence. In summary, every 6 day administration schedule of TMZ or CPA improves survival in orthotopic GL261 GBM with respect to controls or non-metronomic therapy, in partial agreement with previous work on subcutaneous GL261.

Metabolomics of Therapy Response in Preclinical Glioblastoma: A Multi-Slice MRSI-Based Volumetric Analysis for Noninvasive Assessment of Temozolomide Treatment

Glioblastoma (GBM) is the most common aggressive primary brain tumor in adults, with a short survival time even after aggressive therapy. Non-invasive surrogate biomarkers of therapy response may be relevant for improving patient survival. Previous work produced such biomarkers in preclinical GBM using semi-supervised source extraction and single-slice Magnetic Resonance Spectroscopic Imaging (MRSI). Nevertheless, GBMs are heterogeneous and single-slice studies could prevent obtaining relevant information. The purpose of this work was to evaluate whether a multi-slice MRSI approach, acquiring consecutive grids across the tumor, is feasible for preclinical models and may produce additional insight into therapy response. Nosological images were analyzed pixel-by-pixel and a relative responding volume, the Tumor Responding Index (TRI), was defined to quantify response. Heterogeneous response levels were observed and treated animals were ascribed to three arbitrary predefined groups: high response (HR, n = 2), TRI = 68.2 ± 2.8%, intermediate response (IR, n = 6), TRI = 41.1 ± 4.2% and low response (LR, n = 2), TRI = 13.4 ± 14.3%, producing therapy response categorization which had not been fully registered in single-slice studies. Results agreed with the multi-slice approach being feasible and producing an inverse correlation between TRI and Ki67 immunostaining. Additionally, ca. 7-day oscillations of TRI were observed, suggesting that host immune system activation in response to treatment could contribute to the responding patterns detected.

MRSI-based molecular imaging of therapy response to temozolomide in preclinical glioblastoma using source analysis

Characterization of glioblastoma (GB) response to treatment is a key factor for improving patients' survival and prognosis. MRI and magnetic resonance spectroscopic imaging (MRSI) provide morphologic and metabolic profiles of GB but usually fail to produce unequivocal biomarkers of response. The purpose of this work is to provide proof of concept of the ability of a semi-supervised signal source extraction methodology to produce images with robust recognition of response to temozolomide (TMZ) in a preclinical GB model. A total of 38 female C57BL/6 mice were used in this study. The semi-supervised methodology extracted the required sources from a training set consisting of MRSI grids from eight GL261 GBs treated with TMZ, and six control untreated GBs. Three different sources (normal brain parenchyma, actively proliferating GB and GB responding to treatment) were extracted and used for calculating nosologic maps representing the spatial response to treatment. These results were validated with an independent test set (7 control and 17 treated cases) and correlated with histopathology. Major differences between the responder and non-responder sources were mainly related to the resonances of mobile lipids (MLs) and polyunsaturated fatty acids in MLs (0.9, 1.3 and 2.8 ppm). Responding tumors showed significantly lower mitotic (3.3 ± 2.9 versus 14.1 ± 4.2 mitoses/field) and proliferation rates (29.8 ± 10.3 versus 57.8 ± 5.4%) than control untreated cases. The methodology described in this work is able to produce nosological images of response to TMZ in GL261 preclinical GBs and suitably correlates with the histopathological analysis of tumors. A similar strategy could be devised for monitoring response to treatment in patients. Copyright © 2016 John Wiley & Sons, Ltd.

Accurate grading of brain gliomas by soft independent modeling of class analogy based on non-negative matrix factorization of proton magnetic resonance spectra

Hydrogen magnetic resonance spectroscopy ((1) H-MRS) is a non-invasive technique which provides a 'frequency-signal intensity' spectrum of biochemical compounds of tissues in the body. Although this method is currently used in human brain studies, accurate classification of in-vivo (1) H-MRS is a challenging task in the diagnosis of brain tumors. Problems such as overlapping metabolite peaks, incomplete information on background component and low signal-to-noise ratio disturb classification results of this spectroscopic method. This study presents an alternative approach to the soft independent modeling of class analogy (SIMCA) technique, using non-negative matrix factorization (NMF) for dimensionality reduction. In the adopted strategy, the performance of SIMCA was improved by application of a robust algorithm for classification in the presence of noisy measurements. Total of 219 spectra from two databases were taken by water-suppressed short echo-time (1) H-MRS, acquired from different subjects with different stages of glial brain tumors (Grade II (26 cases), grade III (24 cases), grade IV (41 cases), as well as 25 healthy cases). The SIMCA was performed using two approaches: (i) principal component analysis (PCA) and (ii) non-negative matrix factorization (NMF), as a modified approach. Square prediction error was considered to assess the class membership of the external validation set. Finally, several figures of merit such as the correct classification rate (CCR), sensitivity and specificity were calculated. Results of SIMCA based on NMF showed significant improvement in percentage of correctly classified samples, 91.4% versus 83.5% for PCA-based model in an independent test set.

Classification of brain tumours from MR spectra: the INTERPRET collaboration and its outcomes

The INTERPRET project was a multicentre European collaboration, carried out from 2000 to 2002, which developed a decision-support system (DSS) for helping neuroradiologists with no experience of MRS to utilize spectroscopic data for the diagnosis and grading of human brain tumours. INTERPRET gathered a large collection of MR spectra of brain tumours and pseudo-tumoural lesions from seven centres. Consensus acquisition protocols, a standard processing pipeline and strict methods for quality control of the aquired data were put in place. Particular emphasis was placed on ensuring the diagnostic certainty of each case, for which all cases were evaluated by a clinical data validation committee. One outcome of the project is a database of 304 fully validated spectra from brain tumours, pseudotumoural lesions and normal brains, along with their associated images and clinical data, which remains available to the scientific and medical community. The second is the INTERPRET DSS, which has continued to be developed and clinically evaluated since the project ended. We also review here the results of the post-INTERPRET period. We evaluate the results of the studies with the INTERPRET database by other consortia or research groups. A summary of the clinical evaluations that have been performed on the post-INTERPRET DSS versions is also presented. Several have shown that diagnostic certainty can be improved for certain tumour types when the INTERPRET DSS is used in conjunction with conventional radiological image interpretation. About 30 papers concerned with the INTERPRET single-voxel dataset have so far been published. We discuss stengths and weaknesses of the DSS and the lessons learned. Finally we speculate on how the INTERPRET concept might be carried into the future.