Hyperspectral imaging and artificial intelligence to detect oral malignancy - part 1 - automated tissue classification of oral muscle, fat and mucosa using a light-weight 6-layer deep neural network

Deep neural networks can differentiate thyroid pathologies on infrared hyperspectral images

BACKGROUND AND OBJECTIVE

The thyroid is a gland responsible for producing important body hormones. Several pathologies can affect this gland, such as thyroiditis, hypothyroidism, and thyroid cancer. The visual histological analysis of thyroid specimens is a valuable process that enables pathologists to detect diseases with high efficiency, providing the patient with a better prognosis. Existing computer vision systems developed to aid in the analysis of histological samples have limitations in distinguishing pathologies with similar characteristics or samples containing multiple diseases. To overcome this challenge, hyperspectral images are being studied to represent biological samples based on their molecular interaction with light.

METHODS

In this study, we address the acquisition of infrared absorbance spectra from each voxel of histological specimens. This data is then used for the development of a multiclass fully-connected neural network model that discriminates spectral patterns, enabling the classification of voxels as healthy, cancerous, or goiter.

RESULTS

Through experiments using the k-fold cross-validation protocol, we obtained an average accuracy of 93.66 %, a sensitivity of 93.47 %, and a specificity of 96.93 %. Our results demonstrate the feasibility of using infrared hyperspectral imaging to characterize healthy tissue and thyroid pathologies using absorbance measurements. The proposed deep learning model has the potential to improve diagnostic efficiency and enhance patient outcomes.

Diagnostical accuracy of hyperspectral imaging after free flap surgery

Microsurgical free-tissue transfer has been a safe option for tissue reconstruction. This study aimed to analyze the diagnostic accuracy of hyperspectral imaging (HSI) after free-tissue transfer surgery. From January 2017 to October 2019, 42 consecutive free-flap surgeries were performed, and their outcomes were analyzed via HSI. Clinical examination of free-flap perfusion was initially performed. Clinical examination findings were subsequently compared with those of HSI. Potential venous congestion with subsequent necrosis was defined as a tissue hemoglobin index of ≥53%. Student's t-test was used to compare the results of the analysis. The evaluation of sensitivity and specificity for flap failure detection was time dependent using the Fisher's exact test. A p-value of ≤0.05 was considered statistically significant. Microsurgical tissue transfer success rate was 84%. Seven patients presented with venous congestion that caused total flap necrosis. Overall, 124 assessments were made. HSI accurately identified 12 out of 19 pathological images: four as false positive and seven as false negative. The sensitivity and specificity of HSI were 57 and 94%, respectively, compared to those of clinical examination that were 28 and 100%, respectively, within 24 h following tissue transfer. The sensitivity and specificity of HSI were 63 and 96%, respectively, compared to those of clinical examination that were 63 and 100%, respectively, within the first 72 h. A tissue hemoglobin index of ≥53% could predict venous congestion after free-flap surgery. HSI demonstrated higher sensitivity than clinical examination within the first 24 h; however, it was not superior compared to clinical findings within 72 h.

A Comprehensive Review of Recent Advances in Artificial Intelligence for Dentistry E-Health

Artificial intelligence has made substantial progress in medicine. Automated dental imaging interpretation is one of the most prolific areas of research using AI. X-ray and infrared imaging systems have enabled dental clinicians to identify dental diseases since the 1950s. However, the manual process of dental disease assessment is tedious and error-prone when diagnosed by inexperienced dentists. Thus, researchers have employed different advanced computer vision techniques, and machine- and deep-learning models for dental disease diagnoses using X-ray and near-infrared imagery. Despite the notable development of AI in dentistry, certain factors affect the performance of the proposed approaches, including limited data availability, imbalanced classes, and lack of transparency and interpretability. Hence, it is of utmost importance for the research community to formulate suitable approaches, considering the existing challenges and leveraging findings from the existing studies. Based on an extensive literature review, this survey provides a brief overview of X-ray and near-infrared imaging systems. Additionally, a comprehensive insight into challenges faced by researchers in the dental domain has been brought forth in this survey. The article further offers an amalgamative assessment of both performances and methods evaluated on public benchmarks and concludes with ethical considerations and future research avenues.

Spectral Similarity Measures for In Vivo Human Tissue Discrimination Based on Hyperspectral Imaging

PROBLEM

Similarity measures are widely used as an approved method for spectral discrimination or identification with their applications in different areas of scientific research. Even though a range of works have been presented, only a few showed slightly promising results for human tissue, and these were mostly focused on pathological and non-pathological tissue classification.

METHODS

In this work, several spectral similarity measures on hyperspectral (HS) images of in vivo human tissue were evaluated for tissue discrimination purposes. Moreover, we introduced two new hybrid spectral measures, called SID-JM-TAN(SAM) and SID-JM-TAN(SCA). We analyzed spectral signatures obtained from 13 different human tissue types and two different materials (gauze, instruments), collected from HS images of 100 patients during surgeries.

RESULTS

The quantitative results showed the reliable performance of the different similarity measures and the proposed hybrid measures for tissue discrimination purposes. The latter produced higher discrimination values, up to 6.7 times more than the classical spectral similarity measures. Moreover, an application of the similarity measures was presented to support the annotations of the HS images. We showed that the automatic checking of tissue-annotated thyroid and colon tissues was successful in 73% and 60% of the total spectra, respectively. The hybrid measures showed the highest performance. Furthermore, the automatic labeling of wrongly annotated tissues was similar for all measures, with an accuracy of up to 90%.

CONCLUSION

In future work, the proposed spectral similarity measures will be integrated with tools to support physicians in annotations and tissue labeling of HS images.

Comparison of Hyperspectral Imaging and Microvascular Doppler for Perfusion Monitoring of Free Flaps in an In Vivo Rodent Model

To reduce microvascular free flap failure (MFF), monitoring is crucial for the early detection of malperfusion and allows timely salvage. Therefore, the aim of this study was to evaluate hyperspectral imaging (HSI) in comparison to micro-Doppler sonography (MDS) to monitor MFF perfusion in an in vivo rodent model. Bilateral groin flaps were raised on 20 Sprague−Dawley rats. The femoral artery was transected on the trial side and re-anastomosed. Flaps and anastomoses were assessed before, during, and after the period of ischemia every ten minutes for overall 60 min using HSI and MDS. The contralateral sides’ flaps served as controls. Tissue-oxygenation saturation (StO2), near-infrared perfusion index (NPI), hemoglobin (THI), and water distribution (TWI) were assessed by HSI, while blood flow was assessed by MDS. HSI correlates with the MDS signal in the case of sufficient and completely interrupted perfusion. HSI was able to validly and reproducibly detect tissue perfusion status using StO2 and NPI. After 40 min, flap perfusion decreased due to the general aggravation of hemodynamic circulatory situation, which resulted in a significant drop of StO2 (p < 0.005) and NPI (p < 0.005), whereas the Doppler signal remained unchanged. In accordance, HSI might be suitable to detect MFF general complications in an early stage and further decrease MFF failure rates, whereas MDS may only be used for direct complications at the anastomose site.

Trends in hyperspectral imaging: from environmental and health sensing to structure-property and nano-bio interaction studies

Hyperspectral imaging (HSI) is a technique that allows for the simultaneous acquisition of both spatial and spectral information. While HSI has been known for years in the field of remote sensing, for instance in geology, cultural heritage, or food industries, it recently emerged in the fields of nano- and micromaterials as well as bioimaging and -sensing. Herein, the attractiveness of HSI arises from the suitability for generating knowledge about environment-specific optical properties, such as photoluminescence of optical probes in a biological sample or at a single-crystal/particle level, to be leveraged into better understanding of structure-property relationships and nano-bio interactions, respectively. Moreover, given its excellent spectral resolution, HSI is highly suitable for optical multiplexing in multiple dimensions, as sought after for, e.g., high throughput biological imaging by simultaneous tracking of multiple targets. Overall, HSI is an emerging technique that has the potential to transform analytical approaches from biomedicine to advanced materials research. This Trends Article provides insight into the potential of HSI, highlighting selected examples from well-established fields including environmental monitoring and food quality control to set the stage for the discussion of emerging opportunities at the micro- and nanoscale. Herein, special focus is set on photoluminescent micro- and nanoprobes for health and spectral conversion applications.

New Approach to the Old Challenge of Free Flap Monitoring-Hyperspectral Imaging Outperforms Clinical Assessment by Earlier Detection of Perfusion Failure

In reconstructive surgery, free flap failure, especially in complex osteocutaneous reconstructions, represents a significant clinical burden. Therefore, the aim of the presented study was to assess hyperspectral imaging (HSI) for monitoring of free flaps compared to clinical monitoring. In a prospective, non-randomized clinical study, patients with free flap reconstruction of the oro-maxillofacial-complex were included. Monitoring was assessed clinically and by using hyperspectral imaging (TIVITA™ Tissue-System, DiaspectiveVision GmbH, Pepelow, Germany) to determine tissue-oxygen-saturation [StO₂], near-infrared-perfusion-index [NPI], distribution of haemoglobin [THI] and water [TWI], and variance to an adjacent reference area (Δreference). A total of 54 primary and 11 secondary reconstructions were performed including fasciocutaneous and osteocutaneous flaps. Re-exploration was performed in 19 cases. A total of seven complete flap failures occurred, resulting in a 63% salvage rate. Mean time from flap inset to decision making for re-exploration based on clinical assessment was 23.1 ± 21.9 vs. 18.2 ± 19.4 h by the appearance of hyperspectral criteria indicating impaired perfusion (StO₂ ≤ 32% OR StO₂Δreference > −38% OR NPI ≤ 32.9 OR NPIΔreference ≥ −13.4%) resulting in a difference of 4.8 ± 5 h (p < 0.001). HSI seems able to detect perfusion compromise significantly earlier than clinical monitoring. These findings provide an interpretation aid for clinicians to simplify postoperative flap monitoring.