MRS verileri kullanılarak makine öğrenmesi yöntemleri ile MS hastalığının teşhis ve sınıflandırılması = Diagnosis and classification of MS disease with machine learning methods using MRS data

Abstract

Multipl Skleroz (MS), merkezi sinir sistemi (MSS)'nin otoimmun, enflamatuar, demiyelinizan hastalığı olup çoğunlukla genç erişkin çağda görülür. Hastalığın tam nedeni bilinmemekle beraber yapılan araştırmalarda genetik yatkınlık, güneş ışığına fazla maruz kalma, kuzey bölgelerde yaşama ve Epstein-Barr virüsü hastalığın başlıca nedenleri arasında yer almaktadır. MS hastalığı dört tipte sınıflandırılmakta olup en yaygın görülen MS tipleri relapsing-remitting MS (RRMS) ve secondary progressive seyirli MS (SPMS)'dir. MS hastalığının teşhisinde klinik bulgular, beyin omurilik sıvısı incelemeleri, uyarılmış potansiyeller ve manyetik görüntüleme bulguları kullanılmaktadır. Manyetik Rezonans (MR), Multipl Skleroz (MS) hastalığının tanısı ve takibinde kullanılan en önemli görüntüleme tekniğidir. Ancak, MR görüntülemede MS lezyonları beyin tümörlerine benzerlik gösterebilmektedir. Bu nedenle MS tiplerinin sınıflandırılmasında gelişmiş MR tekniklerinden yararlanılmaktadır. Gerçekleştirilen tez çalışmasında Manyetik Rezonans Spektroskopi (MRS) görüntüleme tekniği ve makine öğrenmesi metotlarından faydalanarak kontrol grubu, RRMS ve SPMS tiplerinin otomatik olarak sınıflandırılmasına yönelik bilgisayar destekli tespit sistemi önerilmiştir. Çalışmada kullanılan MRS verileri takip altında olan MS hastalarından bir yıl süresince alınmıştır. Çalışmada birbirinden habersiz iki nöroloji uzmanı tarafından teşhis edilen Kontrol (n=30), RRMS (n=36) ve SPMS (n=25) olmak üzere toplam 91 MS hastası için MRS çekimleri gerçekleştirilmiştir. Hastalardan alınan MRS verilerinden sinyal işleme yöntemleri ile metabolitler elde edilmiştir. Çalışmada yapay bağışıklık sistemi (YBS) tabanlı hibrid bir özellik çıkarım yöntemi önerilmiştir ve bu yöntem mevcut özellik çıkarım yöntemleri olan Peak Integration (PI) ve Principal Component Analysis (PCA) ile kıyaslanmıştır. MRS sinyallerinden çıkarılan özellikler SVM yöntemi ile Kontrol-RRMS ve RRMS-SPMS olmak üzere ikili olarak sınıflandırılmıştır. Gerçekleştirilen deney sonuçlarına göre, önerilen sisteminin kullanımı ile kontrol ve RRMS vakaları birbirlerinden %95 doğruluk ve %90.91 hassasiyet ile ayrılırken, RRMS ve SPMS grupları ise %88.89 doğruluk ve %90.91 hassasiyet ile otomatik olarak sınıflandırılmıştır. Çalışmada önerilen YBS tabanlı hibrit özellik çıkarım metodunun diğer özellik çıkarım metotlarına göre daha başarılı olduğu görülmüştür. Sonuç olarak MRS'nin bilgisayar destekli teşhis yaklaşımı ile birlikte kullanılması, MS seyir tiplerinin sınıflandırılmasında MR ve diğer yöntemleri destekleyici ek bir yöntem olabileceğini göstermektedir.

Multiple sclerosis (MS), is an inflammatory autoimmune disorder of the central nervous system and it is more common in young adult age. The immune system cause damage to myelin formed by cells, called oligodendrocytes, that surround neurons in central nervous system (CNS). MS symptoms can commonly appear between the ages of 20 and 40. MS is 3 times more common in female than male. MS is described as clinically isolated syndrome (CIS), relapsing-remitting multiple sclerosis (RRMS) and progressive multiple sclerosis (PMS). RRMS is characterized as active or non-active. PMS, which can be primary progressive (PP) or secondary progressive (SP), has four possible sub-classifications considering the disability level. Clinical symptoms and findings, cerebrospinal fluid (CSF) examinations, and magnetic resonance imaging (MRI) findings have been used to diagnose MS. The widespread use of MRI has revolutionized the diagnosis and monitoring of MS. Nonetheless, MRI findings of MS types are similar to brain lesions so advanced MRI techniques should be used differentiation of MS types. MR spectroscopy (MRS) is a convenient alternative method to analyze MS, understand its pathogenesis, and determine its course. MRS is an unique MRI technique that measures biochemical changes and metabolites (NAA, choline, creatine etc.) of lesions in the brain. In this study, it has been examined the combination of MRS and the machine learning method for automatic binary classification of Healthy Control Group, RRMS and SPMS types. MS MRS data were obtained from 61 consecutive MS patients who voluntarily participated in this study, which was conducted in the MS Clinic of the Neurology Department at the Bezmialem University Hospital. In addition, a healty control group was created with demographic characteristics similar to those of the RRMS group. Two neurologists with clinical experience in MS and blinded to each other confirmed the diagnosis of healthy controls, RRMS, and SPMS. Among 61 MS patients, 36 were diagnosed with RRMS and the remainder with SPMS. The healthy control group consisted of 30 participants with a similar age to that of the RRMS group and no statistically significant difference. Certain signal processing methods were used to determine the metabolites in the MRS signals. Single-voxel spectroscopy (SVS) data were obtained from Siemens .rda files. SVS raw data were analyzed with the TARQUIN software. TARQUIN is an accurate and robust algorithm for assessing and quantifying single-voxel MRS analysis in the time domain. Time-domain signals were transformed into frequency-domain ones using Fourier transform for the actual quantification. Conventional MRI and SVS data were examined by two radiology experts with at least 10 years of experience in the field. All SVS data were reviewed for quality and assessed with quality control (QC) criteria. Following the experts' opinion, SVS spectra of insufficient quality were not included in the final data set. In addition, all SVS data reached the TARQUIN quality control values. SVS comprises 1024 data points in the TARQUIN software. In this step, SVS data features were extracted, and the most representative ones were determined. In this study, articifial immune system (AIS)-based ensemble feature extraction method is proposed. The artificial immune system was developed inspired by the human immune system. Negative selection algorithm (NSA), clonal selection algorithm (CSA), immune network model algorithm (INME) and danger model immune algorithm (DMIA) are frequently preferred in the AIS. The proposed method mainly consist of training, optimization, windowing, testing, collection and creation of feature vectors. The proposed method is compared with Peak Integration ve Principal Component Analysis methods. The features vectors obtained in the feature extraction step were classified as Healthy Control Group-RRMS and RRMS-SPMS with SVM method. SVM is frequently adopted in fields such as image processing, statistics, and machine learning. This method can classify two or more classes of linear or non-linear data. It counts with optimization techniques, which attempt to find the optimal separating plane between the two classes. The SVM algorithm classifies the features that cannot be separated linearly with Kernel functions. Linear, radial basis, polynomial, and gaussian Kernel functions are commonly used. This study used the quadratic kernel function. Quadratic kernel function is a popular form of polynomial kernel function. In the first evaluation, we assessed the detectability of MS types according to metabolite changes by performing a basic statistical analysis of the dataset. Based on the analysis, the mean levels of NAA peaks were 5.93±2.92, 9.24±2.01, and 7.70±2.85 in healthy controls, RRMS patients, and SPMS patients, respectively. These values may reflect a decreasing trend in NAA peak in progressive forms of MS. In healthy controls, the mean level of Cr and Cho metabolites were 2.93±1.75 and 2.83±1.86, respectively. The mean levels of the Cr and Cho metabolites were 5.88±1.41 and 5.89±1.42, respectively, in RRMS patients and 4.93±1.95 and 4.93±2.11, respectively, in SPMS patients. The metabolite ranges are closer in the RRMS and SPMS groups. Therefore, differentiating MS types with the help of basic statistical methods is difficult. In the second evaluation, healthy controls, RRMS and SPMS patients were categorized in binary classification. The performance of the proposed system, which was developed to overcome the mentioned limitation in the differentiation of MS types, was evaluated according to accuracy, sensitivity and specificity parameters. According to experimental results, the proposed system classify Control Group and RRMS with 95% accuracy, 90.91% sensitivity and 100% specificity, respectively. In addition to this, the dataset were evaluated k-fold (4-fold) cross-validation technique. In this method, the SVS dataset was randomly divided into four parts – one used for the test and the remaining three for training. The total performance of RRMS and healty conrol classification was: accuracy: 90.90%, sensitivity: 91.66% and specificity: 90%. In the second evaluation, RRMS and SPMS patients were categorized in binary classification. Similarly, RRMS and SPMS are classified with 88.89% accuracy, 90.91% sensitivity and specificity: 85.71%. In addition, RRMS and SPMS dataset were evaluated k-fold (4-fold) cross-validation technique. The total performance of RRMS and SPMS classification was: accuracy: 88.52%, sensitivity: 88.89% and specificity: 88%. According to the results, a novel CAD approaches combined with MRS may provide supportive means to MRI for diagnosing and classifying different MS types. We can affirm that SVS associated with machine learning approaches has the potential to contribute further to identifying MS types. Differentiating between healthy controls, RRMS cases, and SPMS cases is clinically important since the type of MS determines the treatment strategy. If the RRMS-SPMS differentiation occurs at a very early stage, the treatment algorithm can be organized accordingly. Some limitations of our study and areas for future research should be mentioned. The most important factor that determined the success of our approach is the training dataset. If the MRS dataset is enriched with healthy control, RRMS, and SPMS samples, the success of our method increases due to better learning of MS cases. Another limitation of the study was obtaining MRS data from a single MR scanner. In future studies, the proposed CAD can be evaluated with MRS data collected from different MR scanners. Moreover, a future study is planned in which RRMS, SPMS, and PPMS will be compared separately with sufficient numbers of patients in each group.