top of page
Writer's picturePre-Collegiate Global Health Review

Increasing the Accuracy of Early Parkinson’s Disease Diagnosis with Slow Saccadic Intrusions

Updated: Jun 24

By Myungha Kim, Dalian American International School, Dalian, Liaoning, China 


Abstract 

Parkinson’s Disease (PD) is a chronic neurodegenerative disorder that causes uncontrollable movements in the body, including tremors and impaired balance. Currently, PD is the second most prevalent neurodegenerative disease, with more than 10 million people living with the disease, and there is no cure. The disease has a nearly 40% seven-year case-fatality rate and while there are treatments that can be used to slow the progression of the disease and control certain symptoms, they may not always be effective. Therefore, early detection of the disease is crucial. However, as the diagnosis of PD heavily relies on subjective physician judgment and rarely on clinical tests, detecting PD at an early stage is often challenging and inaccurate. This study suggests implementing an algorithm based on the implicit piecewise polynomial approximation model for the early diagnosis of PD as it can detect slow saccades with high precision and accuracy. Future studies should further investigate the efficacy of the application of the algorithm in the clinical diagnosis of PD.  

 

Introduction 

Parkinson’s Disease (PD) progresses as dopaminergic neurons located in the midbrain stop functioning properly, resulting in motor system failures (Alexander, 2004).  Early diagnosis of PD can potentially increase the effectiveness of treatments and reduce symptoms like dyskinesia (Murman, 2012). However, PD diagnosis can be difficult as the symptoms can vary widely among individuals. In fact, a study has shown that the accuracy of the early detection of PD is only 58% (Beach and Adler, 2019). 3.6% to 19% of PD diagnosed patients were found to lack evidence of dopaminergic deficit on imaging, indicating that they are not likely to have PD. The inaccuracy of early detection of PD emphasizes the urgent need to develop more effective diagnostic measures. 


Saccadic intrusions are involuntary rapid eye movements that interrupt fixation on a target. Abnormalities of saccades are a common symptom in the early stages of many movement disorders, including PD (Termsarasab, et. al 2015). A recent study reveals that PD patients have slower saccades compared to healthy controls, indicating that detecting the number of slow saccades could be a viable method in detecting PD (Fooken, et. al. 2022). This study puts forth Dai and colleagues’ proposed algorithm based on the implicit piecewise polynomial model to increase the accuracy of detecting slow saccades for early diagnosis of PD (Dai, et. al. 2021). The mean score of the proposed algorithm at detecting saccadic intrusions at different noise levels when compared to previous algorithms for saccade detection is a perfect 1.000. The next most accurate algorithm, Zembly’s algorithm, has a mean score of 0.996.  

 

Algorithm Based on Implicit Piecewise Polynomial Approximation Model 

Dai and colleagues have proposed an algorithm based on the implicit piecewise polynomial approximation model that contains two major steps, including the nonlinear denoising step and basic velocity-threshold (VT) step. The denoising step is used to keep track of abrupt changes that occur with saccadic eye movement and is included as slow saccades produce more noise than normal saccades. This step is unique to this proposed algorithm and increases the accuracy in detecting slow saccades.  


The VT step allows the detection of saccades by applying the VT to the velocity of the estimated time series x. This step shows that the velocity is piecewise linear during saccadic intrusions and zero when fixed. The velocity must exceed 30 degrees/second to be considered a saccade.  

 

Results of the Algorithm 

Figure 1 from the study by Dai and colleagues reproduced below presents how the proposed denoising step successfully improves the reduction of noise in both the position and velocity of the saccades, ensuring a more reliable detection of saccades. 


Figure 1: Noise signals after the proposed denoising step (Dai, et. al. 2021).


Figure 2 reveals that as the noise levels increase, the existing algorithms face more difficulties in detecting slow saccades. Meanwhile, the proposed algorithm almost never misses the detection of slow saccades, showing its high accuracy rate. Table 1 also presents the summary of the proposed algorithm’s high accuracy rate of detecting slow saccades compared to other algorithms. 

Figure 2: Accuracy of true saccades detection among different algorithms (Dai, et. al. 2021).


Table 1: Accuracy in detecting slow saccades under different noise levels using simulated time-series (500 samples/second) (Dai, et. al. 2021).

 

Conclusion 

This study introduced Dai and colleague’s algorithm based on the implicit piecewise polynomial approximation model and explored the relationship between slow saccades and PD. As the proposed algorithm successfully remains highly accurate in its detection of slow saccades under increasing noise levels, the study highlights this algorithm’s strong potential to increase the accuracy of the early diagnosis of PD. Given the global burden and fatality of Parkinson’s Disease and the impact of diagnosing the disease early, this algorithm has the potential to impact many individuals’ lives and change the way we address this disease. 


References


153 views0 comments

Comments


bottom of page