Thathsara Nanayakkara

Mr. Thathsara Nanayakkara

Bionics Engineering, Pukyong National University, Busan, Republic of Korea
orcid
Chartered Engineer | Researcher | International Advisor

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 September 2025 in Sensors
Verified
Journal Article
Open access
PDF file
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara
HHH. M. K. K. M. B. Herath
H. M. K. K. M. B. Herath
HMHADI SEDIGH MALEKROODI
HADI SEDIGH MALEKROODI
NVNuwan Madusanka Vithanage
Nuwan Madusanka Vithanage
MYMyunggi Yi
Myunggi Yi
BLByeong-il Lee
Byeong-il Lee
Parkinson’s disease (PD) impairs balance and gait through neuromotor dysfunction, yet conventional assessments often overlook subtle postural deficits during dynamic tasks. This study evaluated the diagnostic utility of center-of-pressure (CoP) features captured by pressure-sensing insoles during the Timed Up and Go (TUG) test. Using 39 PD and 38 control participants from the recently released open-access WearGait-PD dataset, the authors extracted 144 CoP features spanning positional, dynamic, frequency, and stochastic domains, including per-foot averages and asymmetry indices. Two scenarios were analyzed: the complete TUG and its 3 m walking segment. Model development followed a fixed protocol with a single participant-level 80/20 split; sequential forward selection with five-fold cross-validation optimized the number of features within the training set. Five classifiers were evaluated: SVM-RBF, logistic regression (LR), random forest (RF), k-nearest neighbors (k-NN), and Gaussian naïve Bayes (NB). LR performed best on the held-out test set (accuracy = 0.875, precision = 1.000, recall = 0.750, F1 = 0.857, ROC-AUC = 0.921) using a 23-feature subset. RF and SVM-RBF each achieved 0.812 accuracy. In contrast, applying the identical pipeline to the 3 m walking segment yielded lower performance (best model: k-NN, accuracy = 0.688, F1 = 0.615, ROC–AUC = 0.734), indicating that the multi-phase TUG task captures PD-related balance deficits more effectively than straight walking. All four feature families contributed to classification performance. Dynamic and frequency-domain descriptors, often appearing in both average and asymmetry form, were most consistently selected. These features provided robust magnitude indicators and offered complementary insights into reduced control complexity in PD.

19 October 2025
Verified
Book Chapter
Open access
PDF file
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara
HHH. M. K. K. M. B. Herath
H. M. K. K. M. B. Herath
HMHadi Sedigh Malekroodi
Hadi Sedigh Malekroodi
NVNuwan Madusanka Vithanage
Nuwan Madusanka Vithanage
MYMyunggi Yi
Myunggi Yi
BLByeong-il Lee
Byeong-il Lee
Accurate, real-time identification of gait phases is crucial for clinical assessment, rehabilitation monitoring, and wearable health applications. However, most current smart-insole solutions require high-frequency sampling, increasing power consumption and computational load. To address these limitations, we developed a lightweight, custom-designed smart insole system operating at a low frequency of 5 Hz. The system integrates eight force-sensitive resistors (FSRs) and a tri-axial Inertial Measurement Unit (IMU) per foot. Fourteen healthy participants performed five walking trials on a 10 m. At the same time, synchronized insole sensor data and RGB video recordings were captured, with video annotations serving as the ground truth for gait-phase verification. We systematically evaluated five distinct feature sets using six classical machine learning classifiers with participant-wise cross-validation. Pressure-only features classified using a support vector machine yielded the highest macro-F1 score of 0.915, confirming that low-frequency plantar pressure signals effectively discriminate against gait phases without substantial loss in accuracy. In contrast, IMU-only signals demonstrated significantly lower classification performance, highlighting the limited effectiveness of inertial data at low sampling rates. Additionally, we developed a visual analytics pipeline to enhance interpretability, generating spatial plantar pressure heatmaps and activation-frequency maps that clearly illustrate distinct load patterns for each gait phase. Our findings demonstrate that low-frequency plantar-pressure signals provide sufficient temporal and spatial information for reliable gait-phase detection. This approach offers a practical solution for real-time gait monitoring applications.

24 December 2024
Verified
Book Chapter
Closed access
PDF file
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara
GNG. A. S. Y. Nanayakkara
G. A. S. Y. Nanayakkara
KKK. M. C. M. Karunanayaka
K. M. C. M. Karunanayaka
CKC. P. T. K. Kumara
C. P. T. K. Kumara
NKNalan Karunanayake
Nalan Karunanayake

Verified
Conference Paper
Closed access
HHHitinayake Hmsr
Hitinayake Hmsr
TBT M Basnayaka
T M Basnayaka
DDDmgj Dissanayake
Dmgj Dissanayake
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara

9 March 2024
Verified
Conference Paper
Closed access
KKK M C M Karunanayaka
K M C M Karunanayaka
KFK S D Fernando
K S D Fernando
MGM D C P K Gunathilaka
M D C P K Gunathilaka
GNG A S Y Nanayakkara
G A S Y Nanayakkara
CKC P T K Kumara
C P T K Kumara
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara
To maintain a ship’s stability, early detection of bilges and flooding is crucial. Loss of the stability will lead to risk of life and property on-board ships. In order to identify bilge/flooding sites and ring alarm with an audible and visual warning that specifies the bilge location/area on a monitoring panel, ships are equipped with traditional bilge alarm systems. In the aspect of a war ship, flooding risk is higher than that of a cargo/merchant ship. Due to that, conventional monitoring systems have become inefficient to assure safety of life out at sea. Hence, the proposed system is designed to logically reason many aspects of bilge and flooding to give accurate bilge alarms and support decision making. Fuzzy logic has been found as an emerging technology in the research field to comply this research gap. In the proposed system, sensors have been utilized to acquire input variables in real time. Output is forwarded to the end user after applying Mamdani fuzzy inference to decide the bilge/flooding condition and risk severity of the ship. Center of Gravity (COG) de-fuzzification method has been applied to extract the crisp value of the risk severity as a percentage. The test results have been evaluated after installing on-board a warship. The input and output fuzzy membership functions have been optimized by evaluating the results based on domain expert knowledge. Thus, the determined risk factor is proved to be sound enough to take efficient and effective action against flooding or high bilges.

1 November 2025 in Journal of Imaging
Verified
Journal Article
Open access
PDF file
Thathsara NanayakkaraThathsara Nanayakkara
Thathsara Nanayakkara
HMHADI SEDIGH MALEKROODI
HADI SEDIGH MALEKROODI
JSJae-Uk Sul
Jae-Uk Sul
CNChang-Su Na
Chang-Su Na
MYMyunggi Yi
Myunggi Yi
BLByeong-il Lee
Byeong-il Lee
Reliable acupoint localization is essential for developing artificial intelligence (AI) and extended reality (XR) tools in traditional Korean medicine; however, conventional annotation of 2D images often suffers from inter- and intra-annotator variability. This study presents a low-cost dual-camera imaging system that fuses infrared (IR) and RGB views on a Raspberry Pi 5 platform, incorporating an IR ink pen in conjunction with a 780 nm emitter array to standardize point visibility. Among the tested marking materials, the IR ink showed the highest contrast and visibility under IR illumination, making it the most suitable for acupoint detection. Five feature detectors (SIFT, ORB, KAZE, AKAZE, and BRISK) were evaluated with two matchers (FLANN and BF) to construct representative homography pipelines. Comparative evaluations across multiple camera-to-surface distances revealed that KAZE + FLANN achieved the lowest mean 2D error (1.17 ± 0.70 px) and the lowest mean aspect-aware error (0.08 ± 0.05%) while remaining computationally feasible on the Raspberry Pi 5. In hand-image experiments across multiple postures, the dual-camera registration maintained a mean 2D error below ~3 px and a mean aspect-aware error below ~0.25%, confirming stable and reproducible performance. The proposed framework provides a practical foundation for generating high-quality acupoint datasets, supporting future AI-based localization, XR integration, and automated acupuncture-education systems.