Middle Ear Surgery, Endolymphatic Sac Surgery, Cochlear Implant, Labyrinthectomy, Petrous Apex Lesions, Facial Nerve Surgery, and Lateral Skull Base Tumors, Otolaryngological Diseases Requiring Mastoidectomy
Conditions
Brief summary
Developing a system for artificial intelligence to recognize anatomical landmarks in otolaryngological surgery, enabling real-time tracking of critical temporal bone structures during surgery.
Detailed description
Take the example of the Al recognition and prediction of the incus, external semicircular canal, facial nerve, and facial nerve recess. Within the defined surgical area, annotated data points are utilized to identify and segment the incus and the lateral semicircular canal based on their relative positions and angles concerning the posterior wall of the external auditory canal and the surrounding tissues. Detailed descriptions of the incus and lateral semicircular canal within the surgical area include: Incus: The incus is a small anvil-shaped bone located in the middle ear. It connects to the malleus laterally and the stapes medially. Identifying the incus accurately is crucial due to its proximity to the facial nerve and its involvement in the ossicular chain that transmits sound vibrations. Lateral Semicircular Canal: This is one of the three semicircular canals in the inner ear, oriented horizontally. It is involved in detecting rotational movements of the head. Proper identification is necessary to avoid damaging the canal, which could result in vertigo or balance issues. Input features include further contrast adjustment and localized magnification of images. The enhanced images are classified and localized using the trained model, and the consistency of multiple frames is utilized to determine the final positions of the facial nerve and the facial recess. Statistical analysis is conducted to predict the positions of the facial recess relative to the incus and lateral semicircular canal, providing reference information for surgeons. The system continuously monitors changes in the surgical area, offering dynamic feedback and optimizing the model's accuracy and robustness through incremental training.
Interventions
None listed
Sponsors
Sun Yat-Sen Memorial Hospital of Sun Yat-Sen University
Study design
Observational model
CASE_ONLY
Time perspective
OTHER
Eligibility
Sex/Gender
ALL
Age
6 Months to 100 Years
Healthy volunteers
No
Inclusion criteria
1. Performing otolaryngological surgeries such as cochlear implantation, endolymphatic sac decompression, semicircular canal plugging, and acoustic neuroma surgery using a microscope. 2. Microscopic exposure provides a comprehensive and clear view of the surgical area within the temporal bone.
Exclusion criteria
1. No surgical video recording available. 2. Unclear visualization of the surgical area during microscopic otolaryngological procedures. 3. Incomplete visualization of the entire surgical process. 4. Patients who did not undergo high-resolution temporal bone CT at our hospital or Shenzhen Deep Bay Hospital.
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| otolaryngological diseases requiring mastoidectomy | 1 week from admission to discharge | 1. Performing otolaryngological surgeries such as cochlear implantation, endolymphatic sac decompression, semicircular canal plugging, and acoustic neuroma surgery using a microscope. 2. Microscopic exposure provides a comprehensive and clear view of the surgical area within the temporal bone. |
| Relevant indicators to evaluate the accuracy of the model | From enrollment to the end of the study | 1. The F1 score is the harmonic mean of accuracy and recall, which is used to measure the balance between accuracy and recall of the model. The higher the F1 score, the better the model performance. 2. Kappa value (Cohen's Kappa) is a statistical indicator used to measure classification consistency. It is mainly used to evaluate the degree of consistency between two or more classifiers (including manual classification and model classification), or the classification consistency of the same classifier at different times or under different conditions. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Relevant indicators of other evaluation models | From enrollment to the end of the study | 1. mIOU:mIOU (Mean Intersection-over-Union), that is, mean intersection over union, is an important evaluation index in semantic segmentation tasks. The purpose of semantic segmentation is to classify each pixel in an image into different categories.In medical images, different tissue categories are divided. 2. Precision: The precision rate represents the proportion of all samples predicted by the model to be positive that are actually positive. 3. Recall:The recall rate represents the percentage of all samples that are actually positive that are correctly predicted to be positive 4. Oacc:Overall accuracy refers to the proportion of correctly classified samples in a classification task It is a basic and intuitive indicator for measuring the accuracy of a model or classification method. 5. Loss:The Loss value is an indicator used in the training process of machine learning models to measure the degree of difference between the model's prediction results and the true labels. |
Countries
China
Contacts
Primary ContactHaidi Doctor Yang
Backup ContactJiaqi Doctor Pang
Outcome results
None listed