According to Drug Discovery Today’s recent review, artificial intelligence (AI) and its multiple applications are transforming how scientists go about cancer research.
Authors Bhavya Bhasin and Vaishali Y. Londhe from the Mumbai-based Shobhaben Pratapbhai Patel School of Pharmacy & Technology Management said in the particular paper that tumors and their respective treatments are naturally complicated even though AI is redefining how oncologists approach cancer management.
“The uniqueness of cancers makes the mapping of their progression and early diagnosis difficult,” they wrote.
“Deep learning has been applied successfully to areas that were previously difficult to understand and is setting new standards of cancer care.”
Here are some 5 areas where Bhasin and Londhe are convinced that artificial intelligence is making a considerable impact in oncology:
1. Diagnosing Metastases
Bhasin and Londhe claimed that diagnosing skin cancer usually involves dermoscopic analysis and clinical screening, which is then followed by histopathological and biopsy analysis.
However, some of the latest AI advances have triggered the emergence of an approach that is less time-consuming.
A 2017 research carried out by Esteva et al. and later published by Nature utilized 129,450 skin cancer clinical images in training a convoluted neural network how to spot and categorize cancers.
In turn, the AI was eventually in a position to detect malignancies.
Another team of researchers drawn from Oregon State University utilized deep learning in extracting information from gene expression data, which assisted them in classifying the various types of breast cancer cells, uncovering new breast cancer detection biomarkers.
2. Segmenting Tumors
Analyzing the volume of a tumor marks the next stage in the treatment of cancer that follows immediately after diagnosis, even though traditional techniques utilized by radiologists, including response evaluation criteria in solid tumors (RECIST), are not only slow but can also be inaccurate by 50 percent.
Scientists have leveraged convoluted neural networks in segmenting liver tumors, optic pathway gliomas, and brain tumors to a large degree of accuracy.
In a liver cancer-based study, a team utilized CNNs in segmenting liver tumors, especially in follow-up CTs, keying in a baseline CT scan, defining the CT scan and follow-up scan into the convolutional neural network (CNN) in a bid to attain automated segmentation.
“A major advantage of CNNs over semiautomatic methods is that the need to customize handcrafted features is obviated because of their ability to automatically identify features,” Londhe and Bhasin wrote.
3. Applying Precision Histology
The authors claimed that histomorphology has been “revolutionized,” primarily by precision histology, a form of deep learning.
While diagnostics and pathology have for years depended on accurate H&E-stained slides interpretation – a process that can be unreliable and slow- deep neural networks or DNNs are using algorithms that can help expedite the process.
DNNs have been already been utilized in analyzing skin lesions with the same level of accuracy as practicing dermatologists, breaking down images into pixels and combining them to create reproducible aspects that generate a given type of diagnostic pattern.
“It is likely that DNNs will soon be capable of more accurate analyses based on H&E slides owing to the developments in high-throughput whole-slide scanning technologies,” Londhe and Bhasin said.
“This will also lead to the development of a new biological data pool, which will further aid precision oncology.”
4. Tracking Tumor Development
Deep learning has also been used in tracking tumor development.
Researchers working at Fraunhofer Institute for Medical Image Computing created a deep learning model that upgrades itself and becomes increasingly accurate as it continues reading additional MRIs and CTs.
The software also enables easy comparison of images in a bid to track tumor development, especially between a patient’s clinic visits.
Bhasin and Londhe wrote that the technique would be highly useful in detecting cancers of the spine, ribs, and bone since these tumors are mainly disregarded due to time constraints.
5. Assessing Cancer Stages
Evaluating a patient’s stage of cancer is vital for prognosis, even though the authors claimed that the traditional assessment process “is associated with various limitations.”
As another option, researchers created a prediction model that utilized deep learning in forecasting the patients’ rate of survival, especially those who had experienced a gastrectomy.
“The deep learning-based prognosis detection had a superior prediction ability compared with predictions based on the regular Cox regression,” Londhe and Bhasin wrote.
“It showed that deep learning can provide a more individualized and precise risk-based stratification.”