Contact us

    MrMrsMsDr



    Collection and Use of Personal Information. *

    [Peer-Reviewed Publications] Cancers (Basel) – Automated Gleason Scoring and Tumor Quantification in Prostate Core Needle Biopsy Images Using Deep Neural Networks and Its Comparison with Pathologist-Based Assessment.

    Cancers (Basel)

    2019 Nov 25;11(12):1860. doi: 10.3390/cancers11121860.

    Authors: Han Suk Ryu, Min-Sun Jin, Jeong Hwan Park, Sanghun Lee, Joonyoung Cho, Sangjun Oh, Tae-Yeong Kwak, Junwoo Isaac Woo, Yechan Mun, Sun Woo Kim, Soohyun Hwang, Su-Jin Shin, Hyeyoon Chang

    Abstract

    The Gleason grading system, currently the most powerful prognostic predictor of prostate cancer, is based solely on the tumor’s histological architecture and has high inter-observer variability. We propose an automated Gleason scoring system based on deep neural networks for diagnosis of prostate core needle biopsy samples. To verify its efficacy, the system was trained using 1133 cases of prostate core needle biopsy samples and validated on 700 cases. Further, system-based diagnosis results were compared with reference standards derived from three certified pathologists. In addition, the system’s ability to quantify cancer in terms of tumor length was also evaluated via comparison with pathologist-based measurements. The results showed a substantial diagnostic concordance between the system-grade group classification and the reference standard (0.907 quadratic-weighted Cohen’s kappa coefficient). The system tumor length measurements were also notably closer to the reference standard (correlation coefficient, R = 0.97) than the original hospital diagnoses (R = 0.90). We expect this system to assist pathologists to reduce the probability of over- or under-diagnosis by providing pathologist-level second opinions on the Gleason score when diagnosing prostate biopsy, and to support research on prostate cancer treatment and prognosis by providing reproducible diagnosis based on the consistent standards.

    Keywords: deep neural network; gleason scoring system; prostate cancer; prostate core needle biopsy.

    Deep Bio, Clinical Trial for ‘AI-powered Prostate Cancer Diagnosis’ in Korea

    Seoul, South Korea- Deep Bio announced today that it had officially commenced clinical trial on ‘DeepDx-Prostate,’ a prostate biopsy AI-based diagnostic software. Deep Bio has recently received clinical trial plan approval from the Ministry of Food and Drug Safety for ‘DeepDx-Prostate’ and its clinical validation plan on AI-based diagnosis using prostate biopsy based H&E stained digital images.

    Deep Bio will go through a verification and validation of the AI-based diagnostic software on patients who have undergone prostate biopsy using H&E stained digital images. The clinical trial will be conducted at two academic hospitals (Seoul National University Hospital, Korea University Guro Hospital) and will compare the diagnostic results between clinicians and the DeepDx-Prostate. In addition, the trial will evaluate the effectiveness of the diagnostics using AI-based software.

    Pathologists confirm the diagnosis of prostate cancer through examining the prostate tissue H&E slides under a microscope. While the current examination process is pathologists visually interpreting & measuring the slides, on DeepDx-Prostate, it can analyze and calculate the percentage of cancer in the entire tissue based on pixel level. These analysis results automatically generate in the form of a pathology report. The report includes the patient’s cancer staging based on the Gleason grading system. “DeepDx-Prostate is an AI-software that assists pathologists in making decisions on cancer diagnosis, and it can further help them with selective precision readings,” Deep Bio officials said. Deep Bio is expecting approval as early as this year.

    Deep Bio is pursuing to expand in the global market. In 2018, they had granted compliance on ISO13485:2016, and will soon commence on receiving European CE certification and FDA approval. The founder and CEO of Deep Bio, Kim Sun Woo said, “we are continuously expanding our global network by participating in international conferences such as USCAP, AACR and introducing our products. We plan to speed up the development process in other cancer types, such as breast cancer diagnosis.”
    2019.04.11
    DeepBio

    Deep Bio Inc. ranks 1st in the CAMELYON17 Grand Challenge from post submitting i…

    The Deep Bio research team set the record in the Camelyon Challenge 17’ currently positioning them in first place. The Camelyon17 Grand Challenge is a global competition organized by the Diagnostic Image Analysis Group (DIAG) and the Radboud University Medical Center to evaluate algorithms that perform automated detection of breast cancer metastases in whole-slide images of lymph node sections.

    The Deep Bio research team used their own unique artificial intelligence training method to detect metastasis and predict pN-stage from lymph node histological slides. Their proposed method achieved a kappa score of 0.9570, setting them in first place of the challenge.

    2019.02.12
    DeepBio

    [Peer-Reviewed Publications] JPTM – Artificial Intelligence in Pathology

    Journal of Pathology and Translational Medicine, 2019
    2019;53(1):1-12. DOI: doi.org/10.4132/jptm.2018.12.16

    Authors: Hye Yoon Chang, Chan Kwon Jung1, Junwoo Isaac Woo, Sanghun Lee, Joonyoung Cho, Sun Woo Kim, Tae-Yeong Kwak,

    Abstract

    As in other domains, artificial intelligence is becoming increasingly important in medicine. In particular, deep learning-based pattern recognition methods can advance the field of pathology by incorporating clinical, radiologic, and genomic data to accurately diagnose diseases and predict patient prognoses. In this review, we present an overview of artificial intelligence, the brief history of artificial intelligence in the medical domain, recent advances in artificial intelligence applied to pathology, and future prospects of pathology driven by artificial intelligence.
    Keywords: Artificial intelligence; Deep learning; Pathology; Image analysis

    Deep Bio launches AI-diagnostic software on prostate cancer ..”grants certifications”

    Deep Bio announced today that it had released its artificial intelligence (AI) diagnostic products for the first time in the domestic press. The company’s first AI software product ‘Deep Dx’ detects prostate cancer using H&E stained prostate biopsy WSI images.

    The founder & CEO of Deep Bio, Sun Woo Kim proposed that the company has completed the preparations to enter the domestic approval process by granting ISO13485:2016 (Quality Management System on Medical Device) and KGMP (Korean medical device manufacturing and quality control standard). He additionally commented that the company is officially ready to step into the overseas market.

    ◇ DeepDx “ Cancer detection with 10-seconds..includes an automatic pathology report.”

    Deep Bio ‘DeepDx’ AI-diagnostic software uses digitally scanned prostate biopsy image to diagnose prostate cancer. Despite the size of the high-resolution image, DeepDx can detect cancer on a WSI image and generate an automatic pathology-like report within 10 seconds.

    Kim mentioned, “The pathology reporting guidelines have recommended to particularly include the percentage of Gleason pattern four as it provides clinically significant information on predicting patients prognosis.” The company’s product ‘DeepDx’ can automatically calculate the proportions of each Gleason pattern and further captures the main cancerous areas on images which are included in the report. Deep Bio’s first products were launched at the DPA (Digital Pathology Association) 2018 held in San Diego, USA and will further launch at USCAP & AACR Annual Meeting 2019.

    Deep Bio CEO, Kim said, “Many pathologists are surprised at the accuracy of our products performances, and have shown significant interest.” 2019 is expected to become an important year for Deep Bio and its commercialization to expand globally. Kim added, “We are also considering attracting investment from global venture capital companies, and we will extend our service offerings to multinational pharmaceutical companies.

    For more information:

    http://www.biospectator.com/view/news_view.php?varAtcId=6777

    2018.12.13
    Biospectator

    Deep Bio recieves ISO 13485

    Deep Bio, an AI-powered medical service startup company, today announced that it had granted ISO 13485:2016 certification for its Quality Management System (QMS).

    ISO 13485 is an internationally recognized standard for companies involved in the medical industry to obtain compliance with the quality management system.

    Deep Bio emphasized achieving compliances like ISO 13485 is crucial for them to ensure the products are designed and manufactured with quality and safety for patients. The medical startup company will continue to move forward to maintain its compliance under applicable standards and receive regulatory approvals including the Food and Drug Administration (FDA) to bring their AI-medical device tools closer to the market worldwide.

    2018.09.17

    AI-driven diagnositc company, Deep Bio Inc. closes $5 Million USD in Series A

    On May 28th, AI startup ‘Deep Bio’ succeeded in Series A funding from various venture capital companies in Korea. Deep Bio is a medicalAI company that develops cancer diagnostic software products. While the products are still under development, it is expected to accelerate the commercialization.

    Deep Bio has closed USD 5.6 Million in its Series A round from domestic venture capital companies. The startup has been recognized as an enterprise worth about USD 14 Million (before investment) in the process of attracting investment.

    DTI & Investment, Hyundai Investment, Sejong Venture Partners, Daesung Venture Investment, and MG Investments acquired a total of 296,298 shares of DeepBio’s redeemable convertible preferred stock (RCPS). Some venture capitalists are also considering additional follow-on investments for additional financial support.

    DeepBio was founded in October 2015 by Kim Sun-Woo, a computer engineer graduate from KAIST University. The company is a startup specializing in the production of medical software using deep learning, one of the artificial intelligence technologies.

    The startup company is conducting clinical trials on its medical devices based on its solutions. It is anticipated that the products will be launched in earnest after completing the clinical procedure. The company plans to actively pursue not only in domestic but also overseas bio markets. Deep Bio is regarded as a representative start-up that provides a combination of information & communication technology (ICT) and bio. In the era of the 4th Industrial Revolution, Deep Bio has gained its reputation from investment companies in the essence that contributes value in domestic and foreign markets.

    An industry representative said, “Deep Bio is a promising venture company in the era of the fourth industrial revolution, and if successful, the company will be able to grow and attract attention in overseas markets.” He said.

    For more information:
    https://www.thebell.co.kr/free/content/ArticleView.asp?key=201802280100056660003574&lcode=00

    2018.03.02
    TheBell

    Deep Bio Inc., takes on a challenge to develop software using AI

    “Learning amount ↑ · Error ↓” Utilization as a clinical supporting device ..

    Artificial Intelligence (AI) has now made its headway into the biotech industry. Google is presenting a new technology revolution where systems such as ‘Alpha Go’ are involved in new drug development, diagnosis, and treatment. However, experts in artificial intelligence do not have it easy to commercializing the technology into the biofield. There are no proven business models in which start-ups and small businesses are struggling to establish and can profit from. In this regard, it is worth noting that Deep Bio, formed in 2015., is a company with real experts in artificial intelligence and IT. Majority of the employees have been developing a model through constant communication and collaboration with medical and bio-professionals. Deep Bio has set early goals to help physicians diagnose prostate cancer more accurately and efficiently.

    Taking a look on the process of diagnosing prostate cancer, the first initial step is measuring the prostate-specific antigen (PSA) level based on a blood test. When the PSA level results are abnormal, a prostate biopsy is recommended to examine further for suspicious areas. The prostate tissues collected from the biopsy test are examined under a microscope by a pathologist. Pathologists use the Gleason Scoring system guideline to observe and determine the cancer severity based on the tissue patterns.

    Sun Woo Kim, the founder, and CEO of Deep Bio pointed out the limitations of the pathological diagnosis process. “Even when the same physician analyzes the same tissue again in time, the diagnostic concordance rate is about 80%. Also, there is an inter and intra-observer variability between pathologists interpreting the slides, so therefore, AI-diagnostic software developed by Deep Bio can be a solution to this problem. Deep Bio’s AI-diagnostic software is built by digitally scanning the tissues and using these images to train the model. “The software can learn and store myriad sample images in a short period because deep learning has the unlimited capacity to learn,” Kim said.

    Kim also mentioned the barriers to examining cancer through artificial intelligence. Due to the variability of tissue staining & slide preparation done by different hospitals, it can cause image variations potentially making it uneasy for AI to analyze images. “To solve this problem, we have developed an algorithm that corrects the new image colors according to the image that is the reference point,” Kim said. The company has confirmed that it has the highest accuracy in reading the images when applying the algorithm developed by Deep Bio.

    The company plans to expand its services into the U.S market espousing the shortage of pathologists. There are companies in the U.S. that are developing artificial intelligence in pathology, and one of the notable companies are PathAI, who was founded by Adityco Kosla & Andrew Beck. PathAI is currently collaborating with Philips on breast cancer and other types of diseases.

    The Deep Bio founder emphasized that “AI cannot wholly replace pathologists. Instead, it can help them make better decisions to treat patients.” And further said, “So it is important for us to keep improving our technology by finding the best mechanisms and methods and apply it into our model.”

    Deep Bio’s vision is to develop a company that can significantly contribute to the bio industry. And Kim further said his goal is to build a company that adds value to society.

    For more information:

    http://www.biospectator.com/view/news_view.php?varAtcId=4030

    2017.09.18
    Biospector

    DeepBio, ‘prostate cancer’ detection using AI ..closes seed rounding

    Deep Bio, is a start-up company that develops artificial intelligence (AI) to diagnose prostate cancer. There are diagnostic imagining methods such as X-rays and computed tomography (CT) to detect prostate cancer, but Deep Bio uses H&E stained biopsy digital whole slide images for AI to analyze prostate glands. Artificial Intelligence examines prostate glands based on tissue pattern recognition and the shape of the cell to help reduce diagnostic errors. To ensure cancer is present, a biopsy procedure is done to confirm it. For this reason, Deep Bio can help physicians make accurate decisions. “It can be used to observe and predict patients’ prognosis,” said Kim Sun-woo, Deep Bio Founder & CEO (46), and further mentioned, “it will significantly reduce the diagnostic error rate.”

    Early this year, Deep Bio has granted a total of 1.9 billion Korean Won seed funding from large investment firms in Korea such as Neo Flux. Kim said, “based on my years of engineering experience, I realized the importance of the engineering ability to create high quality that AI can learn.” Deep Bio develops its own uniquely architectured model that enables AI to learn efficiently. Kim stated that his company predominantly attracted the investors by its exclusive technology.

    For more information:
    http://news.mk.co.kr/newsRead.php?year=2017&no=542748

     

    2017.08.13
    MK.매일경제