Author: Deepbio

Five researches in regards to deep learning-based cancer diagnosis and prognosis highlighted

SEOUL, SOUTH KOREA (PRWEB) MARCH 23, 2022 – Deep Bio, a pioneer in medical AI for pathologic cancer diagnostics, announced five research studies at the 2022 United States and Canadian Academy of Pathology (USCAP) held during March 19th through March 24th in Los Angeles. These deep learning-based pathology algorithms demonstrated their potential utility as cancer diagnostic support software.

The studies explore how deep learning can be utilized in cancer diagnosis, prognosis, and prediction of multiple cancer types, including prostate, breast, lungs, etc. In particular, novel research studies such as “Deep Learning-based Automated Detection of Prostate Cancer Lesions in Hematoxylin Only Visualized Images” and “Breast Cancer Survival Analysis through the Extracted Feature from the Prostate Diagnosis Model” drew attention from the USCAP attendees.

“It is especially meaningful for us to share the studies that are conducted for the first time in the field of cancer diagnosis and survival analysis at this year’s USCAP annual meeting. Notably, the fact that our current deep learning-based algorithm, trained with histomorphological features of prostate cancer, accurately predicted the prognosis of breast cancer suggests that the model can be applied to diagnosis, prognosis, and prediction of other types of cancer,” said Tae Yeong Kwak, the CTO of Deep Bio.

Deep Bio continues to focus on not only AI diagnostics but also R&D and presenting the results in various international conferences and events. The company also continues to build its market presence in the global market through overseas digital pathology solution providers in the US, Europe, and India, as well as conducting research cooperation with Stanford Medical School, Harvard Dana-Farber Cancer Center, and other top research institutions in the US.

Sun Woo Kim, the CEO of Deep Bio said, “with growing interest in digital pathology, we are proud to present meaningful research results on deep learning-based cancer diagnostics at the world’s largest pathology society which we’ve participated in every year since 2018. We will continue to endeavor in novel researches using our deep learning technology to bring innovations in cancer medicine.”

Abstracts Presented Included:
Deep Learning-based Automated Detection of Prostate Cancer Lesions in Hematoxylin Only Visualized Images
Considering Uncertainty Improves Deep Learning-based Lung Cancer Subtyping
Ki-67 Index Regression Using Fully Convolutional Regression Network and Cancer Area Segmentation Network
Automatic Histological Grading of Breast Cancer Resection Tissue
Breast Cancer Survival Analysis through the Extracted Feature from the Prostate Diagnosis Model

About Deep Bio 
Deep Bio Inc. is an AI biotech company with in-house expertise in deep learning, pathology, life sciences, and pharmacotherapeutics. As the country’s first to obtain Korea’s MFDS (Ministry of Food and Drug Safety) approval of an AI-based cancer diagnostic support solution, Deep Bio envisions a suite of AI-based IVD SaMDs (In Vitro Diagnostics Software as a Medical Device) for diagnosis and prognosis of multiple cancers. Deep Bio is actively engaged in the research space and participating in ongoing collaborations with top US medical centers. To learn more, visit http://www.deepbio.co.kr.

DeepDx® Prostate is a clinically-validated AI for prostate core needle biopsy tissue image analysis. Whole-slide images (WSIs) of H&E-stained biopsy tissue specimens are analyzed for prostate cancer, Gleason scores, and grade groups. Extensively tested at US CLIA labs (> 500k cores in 2021), DeepDx® Prostate can alleviate the shortage of pathologists and the resultant increase in workload, while reducing diagnostic subjectivity and variability.” To learn more, visit http://www.deepbio.co.kr.

 

 

Join Deep Bio at USCAP 2022

We are excited to finally meet you in person at USCAP 2022!

Explore how Deep Bio’s AI solutions have shaped digital pathology. Learn about the new insights from our studies.

Deep Bio will be exhibiting on-site this year, and we welcome your visit to our booth #427. Let’s meet and discuss the latest updates on digital pathology and AI.

To arrange a meeting and/or schedule a demo session, please send an email to sales@deepbio.co.kr

Visit www.deepbio.co.kr for more information.

 

Poster Presentation schedule:

AI-mediated analysis of prostate biopsies enables faster and better diagnoses

Healthcare systems are struggling with a pathologist shortfall. In the US, the number of active pathologists dropped by 18 percent from 2007 to 2017; in South Korea, 65 pathology residency slots were available in 2020 but only seven graduates enrolled. At the same time, laboratories must manage the ever-growing oncology workload associated with rising cancer rates and increasing calls for early detection – potentially leading to situations that do not make careful diagnoses possible. Deep Bio believes these challenges can be overcome with the power of artificial intelligence (AI) – providing objective and quantified slide analyses and supporting pathologists to make better informed decisions faster and more consistently.

Deep Bio’s first tool, DeepDx® Prostate, analyzes and assesses H&E-stained prostate biopsy slides to detect cancer – analyzing multi-gigapixel whole-slide images in under 30 seconds. At the same time, it provides clinically meaningful information regarding cancer localization and severity, including percentages of each Gleason score and pattern and total tumor and tissue length. This vastly reduces time and effort on the pathologists’ part – improving workflow efficiency and accuracy.

“The software can provide specialist-level support to pathologists.”
The software assists pathologists by highlighting Gleason patterns 3, 4, and 5, where present, with different colored overlays. The gland-level segmentation of these overlays permits pixel-level accuracy of key metrics such as total tumor-to-tissue ratios and Gleason pattern-to-tumor ratios. DeepDx® Prostate has also been validated in a clinical study published in Cancers, yielding high concordance compared with reference standard generated by three board-certified pathologists, with a quadratic-weighted Cohen’s kappa coefficient of 0.907; 99 percent sensitivity; and 94 percent specificity. It demonstrates that the software can provide specialist-level support to pathologists.

DeepDx® Prostate for quality control and research and development empowers pathologists to make more accurate and consistent decisions with faster turnarounds, enabling improved workflows. “This tool is very helpful for quality assurance, enabling pathologists to re-assess areas that they did not annotate originally, but which the AI algorithm indicated as areas of interest,” says Hillel Kahane, Founder of StarPath, New York, USA, and a user of DeepDx® Prostate.

Moreover, Deep Bio is dedicated to developing innovative solutions in the digital healthcare sector and has active collaborations with renowned universities and laboratories around the world. At the 2021 annual meeting of the American Urological Association, researchers from the Stanford University School of Medicine presented two posters featuring analyses by DeepDx® Prostate – providing external validation of the algorithm and utility of the software in measuring concordance in tumor detection between preoperative MRI and prostatectomy histopathology. Results demonstrated high performance of DeepDx® Prostate in diagnosing and grading cancer on prostatectomy specimens. Recently approved by the University of Utah’s Institutional Review Board, the company plans to conduct a prospective study to validate the performance of DeepDx® Prostate with ARUP Laboratories, a leading reference laboratory in the US.

*DeepDx® Prostate is for Research Use Only.
Original article: thepathologist.com/diagnostics/deep-dive-into-high-level-diagnostics

Korea has emerged as a “technology powerhouse” as the nation is seeing new innovative startups that are working on visualizing the once futuristic technology to real life. Korea Biomedical Review met with innovative CEOs from various healthcare startups to hear about their company and what they believe is going on in Korea’s dynamic healthcare startup scene. The first guest is Kim Sun-woo, CEO of Deep Bio. – Ed.

Korea’s artificial intelligence-based medical equipment manufacturers have increased their global presence based on superior technologies.

Many companies use AI platforms to diagnose cancer. However, Deep Bio wants to set it apart by diagnosing prostate cancer using AI for the first time in Korea.

In April last year, Deep Bio became the first company to receive approval from the Ministry of Food and Drug Safety for an AI-based medical device that diagnosis prostate cancer — DeepDx-Prostate. This in-vitro diagnostic software helps doctors diagnose prostate cancer.

Deep Bio was the first AI medical device maker that diagnoses cancer with biopsy images in the pathology department. There are no AI medical devices diagnosing prostate cancer that has received approval from the FDA. However, with a high unmet need in the field, Deep Bio’s technology has already received recognition worldwide, according to the company.

The company has participated in the United States and Canadian Academy of Pathology (USCAP) every year since 2018 and published papers each year. This year, the company has published four articles.

Korea Biomedical Review met with Deep Bio CEO Kim Sun-woo to learn more about DeepDx-Prostate and its plans.

Deep Bio CEO Kim Sun-woo explains how his company’s devices help doctors make quicker and more accurate diagnoses in a recent interview with Korea Biomedical Review at the company’s headquarters in Guro-gu, Southwestern Seoul.

Question: A solution for diagnosing prostate cancer with digital images seems unusual. Please explain how DeepDx Prostate operates.
Answer: To diagnose prostate cancer, hospitals conduct prostate-specific antigen (PSA) testing.

Suppose the patient has a high PSA level and the physician gives an opinion of prostate cancer. In that case, the hospital removes the tissue with a needle biopsy. The pathologist observes the cells to confirm whether it is cancer or not. If it is cancer, they indicate the severity according to the Gleason Score.

However, there is a limit to what pathologists can do due to the size of the data.

DeepDx Prostate, developed by learning the images diagnosed by a pathologist, is prostate cancer diagnosis assistance software.

It analyzes whole slide imaging of digitized H&E-stained prostate needle biopsy specimens to determine the presence of cancer, and the severity of cancer with Gleason scores 3-5.

The company has verified the performance of DeepDx Prostate through various angles, and our product has proven very good performance compared to the reference standards made by pathologists at multiple hospitals, including the Seoul National University Hospital.

Last year, DeepDx Prostate became the first AI medical device in Korea to obtain a grade 3 in vitro diagnostic medical device approval from the Ministry of Food and Drug Safety as auxiliary software for cancer diagnosis through artificial diagnosis.

In developing the software, the company used thousands of sheets at the time of approval, and we are still adding more data.

As of June this year, the company has also added more than 500,000 core images from the U.S. as quality control.

Q: What advantages does DeepDx Prostate have over conventional prostate cancer screening? Are treatment methods and options also provided?
A: Currently, pathologists conduct prostate cancer diagnosis by looking at the tissue under a microscope to find out how much cancer is in the entire tissue.

When they find cancer, they divide it into a score of 1-5 according to the cancer pattern and calculate the Gleason score.

The Gleason score is the sum of the two patterns occupying the most and the next largest portion of cancer tissue. As patterns one and two are similar to normal cells, hospitals usually consider patterns above three significant.

However, the treatment method is still different for patterns 3-5.

Therefore, it is important to measure the Gleason grade accurately. However, unlike X-rays, which have a size of 1000×1000, the pathological image is very large, with an area of ​​150,000×100,000.

Observing this image with the naked eye through a microscope takes a long time and is difficult to distinguish clearly, so different pathologists sometimes diagnose the same cancer tissue differently.

Since the treatment method may be different in each case, there was a high unmet need for an additional diagnostic tool to increase the agreement between pathologists.

The company believes that artificial intelligence can be the solution to these problems.

We are currently conducting a related study with a medical school in the U.S., and the results are coming out that our AI measures more accurately than the naked eye.

In terms of treatment, the company plans to conduct global clinical trials in the future.

Q: Is DeepDx Prostate being used in the US? What is the FDA approval timeline?
A: Currently, several CLIA (Clinical Laboratory Improvement Amendments) laboratories in the United States have been using DeepDx Prostate d for quality control since last year.

Notably, the company has signed a commercial partnership with Lumea, a digital pathology solution company in the U.S., and has provided products since March of this year.

We have also been receiving commercial rights for image use and are using them for product development.

Analyzing pathological images using AI is a new field, and no company has yet received FDA approval.

Therefore, we are preparing the process by carefully organizing the data, and we plan to submit the licensing data to the FDA within this year. If that happens, I think the licensing schedule will become visible sometime next year.

Q: Have you received any feedback from pathologists who have used DeepDx Prostate?
A: In the U.S., DeepDx Prostate provides metrics such as the presence or absence of cancer, the Gleason grade, the percentage of cancer, and the size of the tissue and cancerous tumor.

Many researchers have evaluated this as very useful and reducing the burden of a diagnosis. For example, it can show cancer that occupies a small area of large size of the entire pathology image with a single click, thereby reducing errors that individuals miss while observing with the naked eye.

Q: Are there any additional products under development after DeepDx Prostate?
A: We are developing an AI software for diagnosing breast and bladder cancer.

Notably, for breast cancer, we are developing a product that detects lymph node metastasis. When performing breast cancer surgery, hospitals check for metastases in the nearby sentinel lymph node, and if there are metastases, they remove them altogether.

We are also expanding our product portfolio for prostate cancer.

The company has developed software that detects cancer by analyzing samples collected through transurethral resection of the prostate, performed on patients with benign prostatic hyperplasia at a commercial level, and is preparing for licensing procedures.

Also, the software related to whole-mount specimen analysis is showing good performance, so it is in the process of commercialization.

Q: What are the short- and long-term goals for Deep Bio?
A: The short-term goal is to generate meaningful revenue by successfully commercializing DeepDx Prostate in the U.S. market.

The long-term goal is to provide a companion diagnostic tool for disease prognosis and treatment through pathological image analysis.

We will provide opportunities for high-risk patients through active treatment while helping improve patients’ quality of life by avoiding overtreatment and reducing side effects for low-risk patients.

Original article: www.koreabiomed.com/news/articleView.html?idxno=12071