Performance of the Pentagon Drawing test for the screening of older adults with Alzheimer's dementia

Explainable deep learning approach for extracting cognitive features from hand-drawn images of intersecting pentagons

Hand drawing, which requires multiple neural systems for planning and controlling sequential movements, is a useful cognitive test for older adults. However, the conventional visual assessment of these drawings only captures limited attributes and overlooks subtle details that could help track cognitive states. Here, we utilized a deep-learning model, PentaMind, to examine cognition-related features from hand-drawn images of intersecting pentagons. PentaMind, trained on 13,777 images from 3111 participants in three aging cohorts, explained 23.3% of the variance in the global cognitive scores, 1.92 times more than the conventional rating. This accuracy improvement was due to capturing additional drawing features associated with motor impairments and cerebrovascular pathologies. By systematically modifying the input images, we discovered several important drawing attributes for cognition, including line waviness. Our results demonstrate that deep learning models can extract novel drawing metrics to improve the assessment and monitoring of cognitive decline and dementia in older adults.

Interpretable deep learning approach for extracting cognitive features from hand-drawn images of intersecting pentagons in older adults

Hand drawing involves multiple neural systems for planning and precise control of sequential movements, making it a valuable cognitive test for older adults. However, conventional visual assessment of drawings may not capture intricate nuances that could help track cognitive states. To address this issue, we utilized a deep-learning model, PentaMind, to examine cognition-related features from hand-drawn images of intersecting pentagons. PentaMind, trained on 13,777 images from 3,111 participants in three aging cohorts, explained 23.3% of the variance in global cognitive scores, a comprehensive hour-long cognitive battery. The model’s performance, which was 1.92 times more accurate than conventional visual assessment, significantly improved the detection of cognitive decline. The improvement in accuracy was due to capturing additional drawing features that we found to be associated with motor impairments and cerebrovascular pathologies. By systematically modifying the input images, we discovered several important drawing attributes for cognition, including line waviness. Our results demonstrate that hand-drawn images can provide rich cognitive information, enabling rapid assessment of cognitive decline and suggesting potential clinical implications in dementia.

Practice effect and test-retest reliability of the Mini-Mental State Examination-2 in people with dementia

Background

The Mini-Mental State Examination-Second Edition (MMSE-2) consists of three visions: a brief version (MMSE-2:BV), a standard version (MMSE-2:SV), and an expanded version (MMSE-2: EV). Each version was equipped with alternate forms (blue and red). There was a lack of evidence on the practice effect and test-retest reliability of the three versions of the MMSE-2, limiting its utility in both clinical and research settings. The purpose of this study was to examine the practice effect and test-retest reliability of the MMSE-2 in people with dementia.

Methods

One hundred and twenty participants were enrolled, of which 60 were administered with the blue form twice (i.e., the same-form group, [SF group]) and 60 were administered with the blue form first and then the red form (alternate-form group, [AF group]). The practice effect was evaluated using a paired t-test and Cohen’s d. The test-retest reliability was examined using the intraclass correlation coefficient (ICC).

Results

For the practice effects, in the SF group, no statistically significant differences were found for the MMSE-2:BV and MMSE-2: EV total scores and eight subtests (p = 0.061–1.000), except for the MMSE-2:SV total score (p = 0.029). In the AF group, no statistically significant differences were found for all three versions of the total scores and subtests (p = 0.106–1.000), except for the visual-constructional ability subtest (p = 0.010). Cohen’s d of all three versions’ total scores and subtests were 0.00–0.20 and 0.00–0.26 for SF group and AF group, respectively. For the test-retest reliability, ICC values for all three versions and eight subtests in SF and AF groups were 0.60–0.93 and 0.56–0.93, respectively.

Conclusion

Our results demonstrated that the practice effect could be minimized when alternate forms of the MMSE-2 were used. The MMSE-2 had good to excellent test-retest reliability, except for three subtests (i.e., visual-constructional ability, registration, and recall). Caution should be taken when interpreting the results of visual-constructional ability, registration, and recall subtests of the MMSE-2.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-021-02732-7.

Developing an Image-Based Deep Learning Framework for Automatic Scoring of The Pentagon Drawing Test

BACKGROUND

The Pentagon Drawing Test (PDT) is a common assessment for visuospatial function. Evaluating the PDT by artificial intelligence can improve efficiency and reliability in the big data era. This study aimed to develop a deep learning (DL) framework for automatic scoring of the PDT based on image data.

METHODS

A total of 823 PDT photos were retrospectively collected and preprocessed into black-and-white, square-shape images. Stratified fivefold cross-validation was applied for training and testing. Two strategies based on convolutional neural networks were compared. The first strategy was to perform an image classification task using supervised transfer learning. The second strategy was designed with an object detection model for recognizing the geometric shapes in the figure, followed by a predetermined algorithm to score based on their classes and positions.

RESULTS

On average, the first framework demonstrated 62%accuracy, 62%recall, 65%precision, 63%specificity, and 0.72 area under the receiver operating characteristic curve. This performance was substantially outperformed by the second framework, with averages of 94%, 95%, 93%, 93%, and 0.95, respectively.

CONCLUSION

An image-based DL framework based on the object detection approach may be clinically applicable for automatic scoring of the PDT with high efficiency and reliability. With a limited sample size, transfer learning should be used with caution if the new images are distinct from the previous training data. Partitioning the problem-solving workflow into multiple simple tasks should facilitate model selection, improve performance, and allow comprehensible logic of the DL framework.

Cognitive impairment and its neuroimaging correlates in spinocerebellar ataxia 2

INTRODUCTION

Cognitive impairment (CI) is reported but is poorly explored in spinocerebellar ataxia 2 (SCA2). This study was undertaken to evaluate and classify cognitive impairment in patients with SCA2 and to identify their grey matter (GM) correlates.

METHODS

We evaluated the neurocognitive profile of 35 SCA2 and 30 age-, gender- and education-matched healthy controls using tests for attention, executive functions, learning and memory, language and fluency, and visuomotor constructive ability. Patients were classified into SCA2 with and without CI based on normative data from population and healthy controls. Furthermore, patients with CI were sub-classified based on the number of impaired domains into multi-domain CI (≥3 domains; MDCI) and limited domain CI (≤2 domains; LDCI). The underlying GM changes were identified using voxel based morphometry.

RESULTS

The mean age at onset, duration of disease, and ataxia score was 28.7 ± 8.51 years, 66.7 ± 44.1 months, and 16.1 ± 4.9 points, respectively. CI was present in 71.4% of SCA2 subjects (MDCI: 42.7%; LDCI: 28.5%). Patients with CI had significant atrophy of the posterior cerebellum, sensorimotor cortex, and superior frontal gyrus (FWE p-value <0.05). Patients with MDCI had significant GM atrophy of the angular gyrus compared to LDCI (FWE p-value <0.05).

CONCLUSION

Patients with CI had significant GM involvement of the posterior cerebellum and frontal lobe, suggestive of impairment in the cerebello-fronto-cortical circuitry.

Relationship between Performance on the Mini-Mental State Examination Sub-Items and Activities of Daily Living in Patients with Alzheimer's Disease

Mini-mental state examination (MMSE) subitems provide useful information about the cognitive status of patients with Alzheimer's disease (AD). If the relationship between MMSE subitems and activities of daily living (ADL) can be shown, the performance of sub-items can predict ADL status and may provide useful information for early ADL intervention. Therefore, the purpose of this study was to investigate the relationship between MMSE subitem scores and ADL. The study sample consisted of 718 patients with AD. Logistic regression analysis using the Physical Self-maintenance Scale (PSMS) and Lawton's Instrumental ADL (L-IADL) was performed with each of the subitems as the dependent variables and the MMSE subitem as the independent variable. As a result, the subitems of MMSE, which are strongly related to each item in PSMS differed (e.g., toilet: registration odds ratio 3.00, grooming: naming 3.66). In the case of L-IADL, most items were strongly associated with "writing" (e.g., shopping: odds ratio 4.29, laundry 3.83). In clinical practice, we often focus only on the total MMSE score in patients with AD. However, the relationship between each MMSE subitem and ADL suggested in this study may be useful information that can be linked to ADL care from the performance of the MMSE subitem.

Automatic, Qualitative Scoring of the Interlocking Pentagon Drawing Test (PDT) based on U-Net and Mobile Sensor Data

We implemented a mobile phone application of the pentagon drawing test (PDT), called mPDT, with a novel, automatic, and qualitative scoring method for the application based on U-Net (a convolutional network for biomedical image segmentation) coupled with mobile sensor data obtained with the mPDT. For the scoring protocol, the U-Net was trained with 199 PDT hand-drawn images of 512 × 512 resolution obtained via the mPDT in order to generate a trained model, Deep5, for segmenting a drawn right or left pentagon. The U-Net was also trained with 199 images of 512 × 512 resolution to attain the trained model, DeepLock, for segmenting an interlocking figure. Here, the epochs were iterated until the accuracy was greater than 98% and saturated. The mobile senor data primarily consisted of x and y coordinates, timestamps, and touch-events of all the samples with a 20 ms sampling period. The velocities were then calculated using the primary sensor data. With Deep5, DeepLock, and the sensor data, four parameters were extracted. These included the number of angles (0–4 points), distance/intersection between the two drawn figures (0–4 points), closure/opening of the drawn figure contours (0–2 points), and tremors detected (0–1 points). The parameters gave a scaling of 11 points in total. The performance evaluation for the mPDT included 230 images from subjects and their associated sensor data. The results of the performance test indicated, respectively, a sensitivity, specificity, accuracy, and precision of 97.53%, 92.62%, 94.35%, and 87.78% for the number of angles parameter; 93.10%, 97.90%, 96.09%, and 96.43% for the distance/intersection parameter; 94.03%, 90.63%, 92.61%, and 93.33% for the closure/opening parameter; and 100.00%, 100.00%, 100.00%, and 100.00% for the detected tremor parameter. These results suggest that the mPDT is very robust in differentiating dementia disease subtypes and is able to contribute to clinical practice and field studies.

Screening for Alzheimer's disease in low-educated or illiterate older adults in Brazil: a systematic review

INTRODUCTION

Cognitive screening instruments are influenced by education and/or culture. In Brazil, as illiteracy and low education rates are high, it is necessary to identify the screening tools with the highest diagnostic accuracy for Alzheimer's disease (AD).

OBJECTIVE

To identify the cognitive screening instruments applied in the Brazilian population with greater accuracy, to detect AD in individuals with a low educational level or who are illiterate.

METHODS

Systematic search in SciELO, PubMed and LILACS databases of studies that used cognitive screening tests to detect AD in older Brazilian adults with low or no education.

RESULTS

We found 328 articles and nine met the inclusion criteria. The identified instruments showed adequate or high diagnostic accuracy.

CONCLUSION

For valid cognitive screening it is important to consider sociocultural and educational factors in the interpretation of results. The construction of specific instruments for the low educated or illiterate elderly should better reflect the difficulties of the Brazilian elderly in different regions of the country.

Psychometric properties of Cognitive Instruments in Vascular Dementia and Alzheimer's disease: a neuropsychological study

OBJECTIVES

To describe elderly performance in the Bender Gestalt Test (BGT) and to discriminate its score by using types of errors as comparison among healthy controls, Alzheimer's disease (AD) patients, and vascular dementia (VD) patients.

METHODS

We performed a cross-sectional analysis of 285 elderly individuals of both sexes, all over 60 years old and with more than 1 year of schooling. All participants were assessed through a detailed clinical history, laboratorial tests, neuroimaging, and neuropsychological tests including the BGT, the Cambridge Cognitive Examination (CAMCOG), the Mini-Mental State Examination (MMSE), the Geriatric Depression Scale (GDS), and the Pfeffer Functional Activities Questionnaire (PFAQ). The BGT scores were not used to establish diagnosis.

RESULTS

Mean BGT scores were 3.2 for healthy controls, 7.21 for AD, and 8.04 for VD with statistically significant differences observed between groups (p<0.0001). Logistic regression analysis was used to identify the main risk factors for the diagnostic groups. BGT's scores significantly differentiated the healthy elderly from those with AD (p<0.0001) and VD (p<0.0001), with a higher area under the curve, respectively 0.958 and 0.982. BGT's scores also showed that the AD group presented 12 types of errors. Types of errors evidenced in the execution of this test may be fundamental in clinical practice because it can offer differential diagnoses between senescence and senility.

CONCLUSION

A cut-off point of 4 in the BGT indicated cognitive impairment. BGT thus provides satisfactory and useful psychometric data to investigate elderly individuals.

Diagnostic value of digital clock drawing test in comparison with CERAD neuropsychological battery total score for discrimination of patients in the early course of Alzheimer's disease from healthy individuals

The early detection of cognitive impairment or dementia is in the focus of current research as the amount of cognitively impaired individuals will rise intensely in the next decades due to aging population worldwide. Currently available diagnostic tools to detect mild cognitive impairment (MCI) or dementia are time-consuming, invasive or expensive and not suitable for wide application as required by the high number of people at risk. Thus, a fast, simple and sensitive test is urgently needed to enable an accurate detection of people with cognitive dysfunction and dementia in the earlier stages to initiate specific diagnostic and therapeutic interventions. We examined digital Clock Drawing Test (dCDT) kinematics for their clinical utility in differentiating patients with amnestic MCI (aMCI) or mild Alzheimer’s dementia (mAD) from healthy controls (HCs) and compared it with the diagnostic value of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) neuropsychological battery total score. Data of 381 participants (138 patients with aMCI, 106 patients with mAD and 137 HCs) was analyzed in the present study. All participants performed the clock drawing test (CDT) on a tablet computer and underwent the CERAD test battery and depression screening. CERAD total scores were calculated by subtest summation, excluding MMSE scores. All tablet variables (i.e. time in air, time on surface, total time, velocity, pressure, pressure/velocity relation, strokes per minute, time not painting, pen-up stroke length, pen-up/pen-down relation, and CDT score) during dCDT performance were entered in a forward stepwise logistic regression model to assess, which parameters best discriminated between aMCI or mAD and HC. Receiver operating characteristics (ROC) curves were constructed to visualize the specificity in relation to the sensitivity of dCDT variables against CERAD total scores in categorizing the diagnostic groups. dCDT variables provided a slightly better diagnostic accuracy of 81.5% for discrimination of aMCI from HCs than using CERAD total score (accuracy 77.5%). In aMCI patients with normal CDT scores, both dCDT (accuracy 78.0%) and CERAD total scores (accuracy 76.0%) were equally accurate in discriminating against HCs. Finally, in differentiating patients with mAD from healthy individuals, accuracy of both dCDT (93.0%) and CERAD total scores (92.3%) was excellent. Our findings suggest that dCDT is a suitable screening tool to identify early cognitive dysfunction. Its performance is comparable with the time-consuming established psychometric measure (CERAD test battery).

Accuracy of praxis test from Cambridge Cognitive Examination (CAMCOG) for Alzheimer's disease: a cross-sectional study

BACKGROUND

Praxis impairment may be one of the first symptoms manifested in dementia, primarily in cortical dementia. The Cambridge Cognitive Examination (CAMCOG) evaluates praxis, but little is known about the accuracy of CAMCOG for diagnosing dementia. The aims here were to investigate the accuracy of praxis and its subitems in CAMCOG (constructive, ideomotor and ideational subitems) for diagnosing Alzheimer's disease (AD) among elderly patients.

DESIGN AND SETTING

Cross-sectional study on community-dwelling elderly people.

METHODS

158 elderly patients were evaluated. CAMCOG, Mini-Mental State Examination and Pfeffer Functional Activities Questionnaire were used. ROC curve analysis was used to establish cutoff points.

RESULTS

The total scores for praxis and the constructive subitem presented significant differences (P < 0.0001) between healthy elderly people and AD patients. Stage of dementia (clinical dementia rating, CDR = 0, 1 and 2) showed that total and constructive praxis can be used to classify the stages of dementia (mild and moderate cases), i.e. constructive praxis classified 88% of the patients with mild dementia (P < 0.0001) while total praxis classified 56% with moderate dementia. Comparison of normal controls (NC) and mild dementia cases showed specificity of 71% and sensitivity of 88% (AUC = 0.88; P < 0.0001).

CONCLUSION

Some praxis subtests can have higher predictive diagnostic value for detecting Alzheimer's disease in mild stages (total praxis AUC = 0.858; P < 0.0001; constructive AUC = 0.972; P < 0.0001). Constructive praxis as measured using CAMCOG may contribute towards diagnosing dementia, because occurrence of impairment of praxis may help in recognizing an evolving dementia syndrome.