Deep Learning for Early Scoliosis Detection Using mmWave Radar Gait Data (ScoliRadar-AI)
A Deep Learning-Based Approach for Early Scoliosis Detection Using mmWave Radar-Based Gait Data
Scoliosis is a sideways curvature of the spine that often develops during childhood and adolescence. When detected early, scoliosis can be managed effectively with non-invasive approaches such as bracing and physiotherapy, while late detection frequently leads to surgical intervention. Current screening methods rely on physical examination and X-ray imaging, which exposes children to ionizing radiation and may miss early-stage cases.
This observational study investigates whether millimeter-wave (mmWave) radar, combined with deep learning (a type of artificial intelligence), can detect early signs of scoliosis by analyzing how a child walks. The radar sensor records subtle movement patterns during walking without using cameras and without producing any identifiable images, fully preserving the participant's privacy. No ionizing radiation is involved.
Pediatric participants attending the orthopedic clinic for routine scoliosis evaluation are invited to walk a short distance in front of a mmWave radar sensor. The collected gait recordings are then analyzed using deep learning models, and the results are compared with the participant's standard clinical scoliosis assessment performed by a pediatric orthopedic specialist. The diagnostic performance of the deep learning model is evaluated using sensitivity, specificity, and overall accuracy.
If the approach proves accurate, it could offer a radiation-free, privacy-preserving, and low-cost alternative for early scoliosis screening in schools, primary healthcare centers, and pediatric orthopedic clinics, ultimately supporting earlier diagnosis and reducing the long-term clinical burden of untreated scoliosis.
Studienübersicht
Status
Bedingungen
Intervention / Behandlung
Detaillierte Beschreibung
Background:
Adolescent Idiopathic Scoliosis (AIS) is the most common form of spinal deformity in children, affecting approximately 2-4% of adolescents worldwide. Early detection is critical because mild curves can often be managed conservatively (bracing, targeted physiotherapy), whereas advanced curves frequently require surgical correction. Current screening primarily relies on physical examination (forward bend test, scoliometer) supplemented by radiographic confirmation. These methods have known limitations: physical examination has variable sensitivity and inter-observer reliability, while repeated radiographic follow-up exposes pediatric patients to cumulative ionizing radiation. Camera-based motion analysis systems have been proposed as alternatives but raise significant privacy concerns in pediatric populations.
Rationale:
Millimeter-wave (mmWave) radar is a non-ionizing, contactless sensing technology that captures fine-grained motion signatures without producing identifiable visual images. Recent advances in deep learning have demonstrated promising results in interpreting radar-derived gait signals for biomechanical analysis. The investigators hypothesize that subtle biomechanical asymmetries associated with early scoliosis can be detected from mmWave radar gait recordings using appropriately trained deep learning models, providing a privacy-preserving and radiation-free screening modality.
Primary Objective:
To develop and evaluate the diagnostic accuracy of a deep learning model that classifies pediatric participants as having scoliosis or not based on mmWave radar gait data, measured by sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC).
Secondary Objectives:
- To compare diagnostic performance across multiple deep learning architectures (including convolutional neural networks, recurrent neural networks, and transformer-based models).
- To evaluate model performance stratified by age, sex, and curve severity (Cobb angle category).
- To assess test-retest reliability of radar-derived gait features.
Study Design:
This is a single-center, prospective, observational diagnostic accuracy study. Pediatric participants undergoing routine scoliosis evaluation at the participating center are invited to take part. Each participant performs a standardized walking task along a defined path in front of a mmWave radar sensor. Radar recordings are processed and analyzed using deep learning models. Model outputs are compared against the reference standard.
Reference Standard:
Scoliosis status is established by a pediatric orthopedic specialist based on clinical examination supplemented by Cobb angle measurement from existing standard-of-care radiographic data. No additional radiographic imaging is performed for the purpose of this study.
Data Handling and Privacy:
All radar recordings and clinical data are de-identified at the point of collection and stored on institutional servers in compliance with the Turkish Personal Data Protection Law (Law No. 6698, KVKK) and the Regulation on Personal Health Data. Data access is restricted to authorized study personnel. No identifiable visual images are recorded by the mmWave radar sensor.
Studientyp
Einschreibung (Geschätzt)
Kontakte und Standorte
Studienkontakt
- Name: Zehra Bilici, MSc
- Telefonnummer: +905343056166
- E-Mail: zbilici@gtu.edu.tr
Studieren Sie die Kontaktsicherung
- Name: Ercan Ayaz, Doç Dr.
Studienorte
-
-
Istanbul
-
Istanbul, Istanbul, Türkei (türkiye), 34480
- Başakşehir Çam and Sakura City Hospital
-
Kontakt:
- Zehra Bilici, MSc
- Telefonnummer: 05343056166
- E-Mail: zbilici@gtu.edu.tr
-
Kontakt:
- Ercan Ayaz, Doç. Dr.
-
-
Teilnahmekriterien
Zulassungskriterien
Studienberechtigtes Alter
- Kind
- Erwachsene
- Älterer Erwachsener
Akzeptiert gesunde Freiwillige
Probenahmeverfahren
Studienpopulation
Beschreibung
Participation Criteria:
- Being between 2 and 75 years of age at the time of registration
- Having applied to the pediatric orthopedics outpatient clinic for an assessment of suspected or known scoliosis
- Being able to walk independently for at least 7 meters without assistive devices
- Written informed consent from a parent or legal guardian
- Written informed consent from the participant
Exclusion Criteria:
- Severe scoliosis requiring urgent surgical intervention that prevents participation in walking tasks
- Refusal to give informed consent or consent
Studienplan
Wie ist die Studie aufgebaut?
Designdetails
Kohorten und Interventionen
Gruppe / Kohorte |
Intervention / Behandlung |
|---|---|
|
Pediatric Participants Undergoing Scoliosis Evaluation
Consecutive pediatric participants attending the orthopedic outpatient clinic for routine scoliosis evaluation.
The cohort includes participants across the full spectrum of clinical assessment outcomes (both scoliosis confirmed and scoliosis ruled out) to enable evaluation of the diagnostic accuracy of the mmWave radar-based deep learning model against the standard-of-care clinical and radiographic reference assessment.
|
Each participant performs a standardized walking task along a defined path in front of a millimeter-wave (mmWave) radar sensor.
The radar continuously records the participant's gait micro-Doppler signatures during the walk.
The mmWave radar device is contactless, non-ionizing, and does not capture identifiable visual images, fully preserving participant privacy.
The recorded gait signals are subsequently processed and analyzed using deep learning models (including convolutional and transformer-based architectures) trained to classify scoliosis status.
The full radar-based assessment takes approximately 5 to 10 minutes per participant.
The standard clinical and radiographic scoliosis evaluation performed as part of routine care serves as the reference standard.
|
Was misst die Studie?
Primäre Ergebnismessungen
Ergebnis Maßnahme |
Maßnahmenbeschreibung |
Zeitfenster |
|---|---|---|
|
Diagnostic Accuracy of the mmWave Radar-Based Deep Learning Model for Scoliosis Detection (AUC-ROC)
Zeitfenster: Assessed at the end of the data collection period, approximately 18 months after study start
|
The diagnostic performance of the mmWave radar-based deep learning classification model is assessed by the area under the receiver operating characteristic curve (AUC-ROC).
The AUC-ROC is computed by comparing the model's predicted probability of scoliosis for each participant against the reference standard (clinical examination combined with Cobb angle measurement from standard-of-care radiographic imaging) on a held-out test set.
The AUC-ROC is reported as a single numeric value between 0 and 1, with 95% confidence intervals.
|
Assessed at the end of the data collection period, approximately 18 months after study start
|
Sekundäre Ergebnismessungen
Ergebnis Maßnahme |
Maßnahmenbeschreibung |
Zeitfenster |
|---|---|---|
|
Sensitivity and Specificity of the Deep Learning Model at the Optimal Operating Point
Zeitfenster: Assessed at the end of the data collection period, approximately 18 months after study start
|
At the operating point that maximizes the Youden index on the validation set, the sensitivity (true positive rate) and specificity (true negative rate) of the deep learning model for classifying scoliosis status are calculated on the held-out test set.
Both metrics are reported as percentages with 95% confidence intervals.
|
Assessed at the end of the data collection period, approximately 18 months after study start
|
|
Comparative Diagnostic Performance Across Deep Learning Architectures
Zeitfenster: Assessed at the end of the data collection period, approximately 18 months after study start
|
The diagnostic accuracy (AUC-ROC) of multiple deep learning architectures, including convolutional neural networks (CNN), recurrent neural networks (RNN/LSTM), and transformer-based models, is compared on the same dataset using cross-validation.
The architecture yielding the highest AUC-ROC is identified as the best-performing model.
|
Assessed at the end of the data collection period, approximately 18 months after study start
|
|
Stratified Diagnostic Performance by Scoliosis Severity (Cobb Angle Category)
Zeitfenster: Assessed at the end of the data collection period, approximately 18 months after study start
|
The diagnostic accuracy of the deep learning model is evaluated separately for participants with mild scoliosis (Cobb angle 10 to 24 degrees), moderate scoliosis (Cobb angle 25 to 39 degrees), and severe scoliosis (Cobb angle 40 degrees or greater).
Sensitivity is reported for each severity category to assess whether the model detects clinically significant curves of different magnitudes.
|
Assessed at the end of the data collection period, approximately 18 months after study start
|
Mitarbeiter und Ermittler
Sponsor
Studienaufzeichnungsdaten
Haupttermine studieren
Studienbeginn (Geschätzt)
Primärer Abschluss (Geschätzt)
Studienabschluss (Geschätzt)
Studienanmeldedaten
Zuerst eingereicht
Zuerst eingereicht, das die QC-Kriterien erfüllt hat
Zuerst gepostet (Tatsächlich)
Studienaufzeichnungsaktualisierungen
Letztes Update gepostet (Tatsächlich)
Letztes eingereichtes Update, das die QC-Kriterien erfüllt
Zuletzt verifiziert
Mehr Informationen
Begriffe im Zusammenhang mit dieser Studie
Schlüsselwörter
Zusätzliche relevante MeSH-Bedingungen
Andere Studien-ID-Nummern
- GTU-Scoliosis-mmWave-2026
Plan für individuelle Teilnehmerdaten (IPD)
Planen Sie, individuelle Teilnehmerdaten (IPD) zu teilen?
Beschreibung des IPD-Plans
Arzneimittel- und Geräteinformationen, Studienunterlagen
Studiert ein von der US-amerikanischen FDA reguliertes Arzneimittelprodukt
Studiert ein von der US-amerikanischen FDA reguliertes Geräteprodukt
Diese Informationen wurden ohne Änderungen direkt von der Website clinicaltrials.gov abgerufen. Wenn Sie Ihre Studiendaten ändern, entfernen oder aktualisieren möchten, wenden Sie sich bitte an register@clinicaltrials.gov. Sobald eine Änderung auf clinicaltrials.gov implementiert wird, wird diese automatisch auch auf unserer Website aktualisiert .