Researchers have developed a machine learning algorithm (MLA) that can diagnose thyroid cancer by analyzing human thyroid cell clusters. The MLA uses both Papanicolaou staining and intrinsic refractive index (RI) imaging contrasts to classify benign and malignant cell clusters. The study analyzed thyroid fine-needle aspiration biopsy (FNAB) specimens from 124 patients, using correlative optical diffraction tomography to measure both color images and three-dimensional RI distribution.
The MLA was able to classify benign and malignant cell clusters using color images, RI images, or both, achieving accuracies of 98.0%, 98.0%, and 100%, respectively. The nucleus size was mainly used in the color image for classification, but the detailed morphological information of the nucleus was used in the RI image. The researchers believe that the complementary information from color and RI images can improve MLA performance and have the potential to diagnose thyroid cancer.
Thyroid nodules are common in the general population, and thyroid fine-needle aspiration biopsy is the most important preoperative diagnostic modality for distinguishing between benign and malignant thyroid nodules. The detection of thyroid nodules and the frequency of thyroid FNAB has increased significantly worldwide with the increasing utilization of diagnostic imaging modalities. However, evaluation of FNABs is still hindered by challenges, including dependence on highly skilled cytopathologists and interobserver variability, which is further complicated by the quality of image data presented for interpretation.
Machine learning algorithms are increasingly being applied to medical imaging and tumor pathology. Recently, MLAs have shown high overall accuracy in diagnosing thyroid cancer using digital imaging of thyroid FNAB specimens. Nevertheless, MLA-based diagnostic tools for thyroid FNAB have not yet been commercialized. The method for acquiring digital data and the retrieval of imaging information to be utilized for MLAs from thyroid FNAB specimens has been poorly studied and neglected, despite being an important determinant of MLA performance.
The study aimed to pursue a higher content of cytopathology end-points and evaluated the potential of diagnoses using standard thyroid FNAB brightfield microscopy images combined with an emerging quantitative phase imaging technique (QPI). QPI exploits the intrinsic refractive index (RI) distribution of cells and tissues as quantitative label-free imaging contrast. RI images can show complementary and synergistic features to brightfield microscope-based color images for the same cells or tissues due to the differences in imaging methods.
Overall, the study has shown promising results in the development of machine learning algorithms for thyroid FNAB diagnoses, with the potential to enhance diagnostic accuracy and reduce both the time required for diagnoses by experts and interobserver variability associated with the quality of image data presented for interpretation.
