Multidimensional classification of hippocampal shape features discriminates Alzheimer's disease and mild cognitive impairment from normal aging (1707.05961v1)

Abstract: We describe a new method to automatically discriminate between patients with Alzheimer's disease (AD) or mild cognitive impairment (MCI) and elderly controls, based on multidimensional classification of hippocampal shape features. This approach uses spherical harmonics (SPHARM) coefficients to model the shape of the hippocampi, which are segmented from magnetic resonance images (MRI) using a fully automatic method that we previously developed. SPHARM coefficients are used as features in a classification procedure based on support vector machines (SVM). The most relevant features for classification are selected using a bagging strategy. We evaluate the accuracy of our method in a group of 23 patients with AD (10 males, 13 females, age $\pm$ standard-deviation (SD) = 73 $\pm$ 6 years, mini-mental score (MMS) = 24.4 $\pm$ 2.8), 23 patients with amnestic MCI (10 males, 13 females, age $\pm$ SD = 74 $\pm$ 8 years, MMS = 27.3 $\pm$ 1.4) and 25 elderly healthy controls (13 males, 12 females, age $\pm$ SD = 64 $\pm$ 8 years), using leave-one-out cross-validation. For AD vs controls, we obtain a correct classification rate of 94%, a sensitivity of 96%, and a specificity of 92%. For MCI vs controls, we obtain a classification rate of 83%, a sensitivity of 83%, and a specificity of 84%. This accuracy is superior to that of hippocampal volumetry and is comparable to recently published SVM-based whole-brain classification methods, which relied on a different strategy. This new method may become a useful tool to assist in the diagnosis of Alzheimer's disease.

• The paper presents a novel method using multidimensional hippocampal shape features classified by support vector machines (SVM) to distinguish Alzheimer's disease and mild cognitive impairment from normal aging.
• The method achieved high accuracy (94% for AD vs controls, 83% for MCI vs controls), significantly outperforming traditional hippocampal volume analysis.
• This fully automated approach has significant potential for streamlining clinical workflows and improving early diagnosis, particularly in identifying prodromal AD (MCI).

Multidimensional Classification of Hippocampal Shape Features in Alzheimer's Disease and Mild Cognitive Impairment

The paper presents a novel method for the discrimination of Alzheimer's Disease (AD) and Mild Cognitive Impairment (MCI) from normal aging, focusing on advanced image processing and machine learning techniques. Specifically, it leverages hippocampal shape features, analyzed through spherical harmonics (SPHARM) and classified using support vector machines (SVM), to distinguish between these conditions. The method outperforms traditional hippocampal volumetry, achieving a classification accuracy of 94% for AD versus controls and 83% for MCI versus controls, with substantial sensitivity and specificity. These metrics underscore the potential of detailed hippocampal morphology in early AD diagnosis.

Methodological Approach

The methodology combines SPHARM with SVM in a multidimensional framework. SPHARM coefficients, used to capture hippocampal shape, serve as inputs for an SVM classifier. The segmentation of the hippocampus from MRI scans was accomplished using a fully automatic method previously validation, providing a solid foundation for feature extraction.

The dimensionality of the data is managed with univariate feature selection, employing a bagging strategy to enhance robustness. In the classification task, leave-one-out cross-validation is implemented to estimate accuracy, addressing concerns regarding model generalizability on a modestly sized data set.

Results and Comparative Analysis

The method's results indicate superior classification performance, notably surpassing previous efforts focused solely on volumetric analysis. For example, manual hippocampal volumetry has been reported with a significantly lower accuracy for MCI classification, ranging from 60% to 74%. The multidimensional shape analysis enhances discriminatory power, indicating that local shape features are more sensitive to early-stage morphological changes than global volumetric measures.

When compared with voxel-based SVM approaches, similar accuracies were achieved for AD classification, but notably higher for MCI classification, indicating the strength of detailed hippocampal shape analysis. The high classification rate demonstrates the effectiveness of focusing on one key structure affected early in AD pathology.

Implications and Future Directions

The implications of this research are significant for improving diagnostic techniques in neurodegenerative disease, particularly given the fully automated nature of the method, which addresses the labor and expertise required in manual segmentation approaches. The automatic segmentation of the hippocampus combined with machine learning could streamline clinical workflows and enhance early diagnosis.

Further, the ability of the method to classify MCI with high accuracy suggests its utility in identifying prodromal AD patients, a crucial aspect for early intervention strategies. Future developments might incorporate this approach into larger integrated classification frameworks that combine hippocampal shape analysis with whole-brain data, potentially refining the specificity and sensitivity across neurodegenerative spectra.

In summary, the paper provides a rigorous examination of hippocampal morphology's role in early-stage Alzheimer's and mild cognitive impairment diagnosis, yielding results that are of substantial utility in clinical neuroimaging and offering a robust alternative to traditional methods. Continued exploration in this vein may lead to more comprehensive models that can effectively augment current AD diagnostic protocols.