Technological integration has evolved the medical imaging science from 2 dimensional to three-dimensional imaging and in turn has enhanced the patient care levels. The past phase at which the healthcare field is evolving keeping up with recent technologies and terminologies itself has become a herculean task. This article is one such effort to simplify and explain about the different advanced imaging technologies that are being integrated into medical imaging at present and their application in oral maxillofacial conditions.
Radiomics, radio-genomics and molecular imaging integrates medical science with engineering, chemistry, molecular biology and computer science.
Molecular imaging involves visualisation and measurement of physiological or pathological process in the living organism at the cellular or molecular level in real time in a non-invasive manner. In here a signal producing imaging agent is introduced into the body and an imaging device capable of detecting the probe’s signal creates detailed images. X-ray computed tomography imaging (CT), optical imaging (OI), radionuclide imaging (involving PET and SPECT), ultrasound (US) imaging and
magnetic resonance imaging (MRI) are some of the imaging modalities capable of measuring the signal produced by the imaging agent.
The imaging agent used can be a radiopharmaceutical or a probes like light, sound. In case of MR spectroscopy, the differences in magnetism is used to measure the changes in body without the use of a probe.
The imaging data generated by advanced modalities like CT, OI, and MRI are under-utilised by the human eyes owing to their high resolution. Radiomics converts this image data into a mathematical output (features) thereby providing additional data for the detection of a condition. The additional information provided is known to be different than those the human eye can resolve from standard images.
For example, from a given tumour imaging a human eye can detect a tumour. But, radiomics can help in diagnosing intra tumoral complexity also from the same set of images.
Radio genomics involves association of the quantitative radiomics data with genome or molecular parameters to correlate it with a clinical outcome or clinical measures.
Both radiomics and radio genomics uses artificial intelligence and deep-learning algorithms to analyse conventional CT or MRI to identify imaging features that may be clinically relevant in patients that are not commonly detected by human eyes.
The work flow for radiomics include 4 stages.
(1) Image acquisition
The first step in radiomics work flow acquiring images for patient diagnosis and disease staging using standard imaging protocols.
(2) Lesion annotation and segmentation of images
This step involves annotating human observations and identifying the tumour region manually. As image feature extraction depends on the specification of the tissue volume in the image to be analysed is decided in this step this is considered the most challenging step in radiomics workflow.
(3) Image feature extraction, selection, and classification
Further the images are converted to quantitative data output forms using suitable algorithms and classified into algorithm groups that generate data on
umor shape, margin sharpness, and texture features. Hundreds of features can be statistically extracted from images within each type. Feature selection allows limiting the number of quantitative image parameters for building radiomics models, which helps to avoid model overfitting in small sample size data.
(4) Model building and outcome prediction
Based on the image feature selection and classification in step three a radiomic model is generated which provides interpretation for patient outcome prediction or risk classification of tumour biologic aggressiveness.
Non -invasiveness of these modalities allows repeated studies in the same animal/patient, thus making it possible to collect longitudinal data and reduce the number of animals and cost. Therefore, molecular imaging plays an important role in earlier detection, accurate diagnosis, and drug development and discovery.
The use of modalities like MR spectroscopy and Raman spectroscopy reduces patient exposure to radiation.
key problems regarding theory, technology, and system, especially molecular imaging agents and imaging equipment, are not solved yet. The probes used in these imaging modalities also need to address several issues like barriers in delivery, biological compatibility, and the diversity between species and multimodal contrast agent fusion etc.
Data generated by radiomics and radiogenomics and molecular imaging depends on images collected by imaging instruments like CT, MRI etc. All these instruments have their own set of limitations. For instance, MRI has poor sensitivity and CT has greater artefacts.
Radiomics is vulnerable to the effects of dental artifacts (DA) caused by metal implants or fillings and can hamper future reproducibility on new datasets
Even though multimodality imaging like MET/MRI or PET/CT has many advantages its still in its earlier stages and needs to find solution for difficulties like designing a PET/MRI system suited for the entire body, cost increase, performance improvement.
Another important challenge in case of radiomics and radiogenomics is the availability of big data concerning dental images. High variability of imaging protocols across the globe can impede comparisons among different studies, and the development of multi-institutional large databases and in turn impact the establishment of the gold standard for validating classifications or predictive models resulted from AI studies.
A further issue that needs to be addressed is the computational speed; which needs to be clinically realistic to enable its use in clinical practice.
Radiomics, radio genomics, and molecular imaging have generally existed as separate silos. However, combining these modalities has resulted in synergistic benefit to patients. For example, integration of imaging modalities like MRI/CT/USG to PET has improved the consistency and effectiveness of cancer detection, staging and assessment of treatment response. Similarly, application of radiomics and radio genomics to these set of images can provide
the diagnostician a more accurate ability to diagnose and predict the treatment outcomes.
For example, Raman spectroscopy as a means of diagnosing dental caries has long been studied. But its application in caries detection was considered difficult as sites of dental caries generate fluorescence noise that exceeds the Raman light and varies according to measurement site and configuration resulting in inaccuracies in diagnosis. However, the combination of Raman spectroscopy with optical coherence tomography and fluorescence subtraction methods using multichannel lock-in detection has been reported to avoid fluorescence interference.
Dr. Anisha Yaji
Senior Scientific Manager, KROYNAS Pvt. Ltd