CAD4COVID projects

Bernhoven Hospital in the Netherlands has seen one of the highest cases of COVID-19 in the country. It is located in an area that has the highest burden of cases nationally. Bernhoven hospital has been successfully using CAD4COVID for the past weeks. In a triage tent set in front of the hospital, people were selected for radiological examination based on their symptoms (e.g. cough, fever). With abnormal CXR results, patients qualified for a RT-PCR test. There was a high prevalence: 90% of the patients had COVID-19 symptoms. In the 60-70% of the cases COVID-19 abnormalities were visible on the CXR. Some patients had COVID-19 abnormalities, but negative RT-PCR results.

In Mexico, there are not much RT-PCR tests available to test patients on COVID-19. They rely on the clinical symptoms and imaging for diagnosing COVID-19. The Hospital Angeles in Ciudad de Mexico now uses the CAD4COVID software on X-ray and CT to screen COVID-19.

The doctors adjust their COVID-19 screening methods to the patient. Dr. Winston Wong from the Department of Radiology tells us:

"For our patients, with respiratory symptoms, as cough, fever, dyspnea, we take a chest X-ray. When their doctor asks for the CT, we do it.
In E.R, patients with respiratory symptoms, we do chest X-ray and chest CT.
For hospitalized patients, we do chest X-ray and sometimes chest CT.
When the cases are on the PACS system, I usually use your software to take note on the CAD4COVID probability score for chest X-ray."

"The first patient with COVID-19 was admitted to the Diospi Suyana hospital. He had travelled some distance: 1200km on his way from Lima to Quillabamba. Passing through the village of Curahuasi, he stopped breathing. The staff of the local health station took a COVID quick test with a positive result. The staff of the state health system brought him to the missionary hospital. His saturation was 75% and he was gasping for air. Within seconds MRI-nurse, Melanie Friesen pushed the mobile X-ray machine onto the isolation ward. The pictures showed the characteristic alterations of the lung. Analysing the pictures with special software from Holland (CAD4COVID) confirmed the diagnosis. Now the patient is on oxygen and is getting all medications as prescribed by the protocol."
Full article

We have asked Hospital Complejo de Albacete what their findings are regarding CAD4COVID:

"CAD4COVID has allowed us to review X-ray images in a standardized way without overloading the activity of radiologists. We will also use CAD4COVID to revise all patients that attended the emergency room and retrospectively compare "human" diagnosis vs your machine learning system. Our radiologists have been able to have an enhanced focus on the COVID-19 patients in the middle of a pandemic episode thanks to the CAD4COVID platform."

The Emergency Radiology Department of Medical Imaging Centre at Semmelweis University is responsible for all the radiological examinations of COVID and non-COVID treatment facilities. The designated centre at the University is the most interlinked with the Emergency Department where the COVID/nonCOVID triage takes place. They provide the imaging for this triage as well as the needs of the intermediary and confirmed COVID treatment wards.

Currently their protocol uses CXR imaging for stable or low risk COVID suspect patients. This investigation arm precedes CT as well as RT-PCR. In most cases CT is reserved for the seriously ill or high risk patients and patients who have ambiguous CXR findings. CXR is the main follow up modality for serious cases whom are treated in the ICU.

"Therefore, we are planning to utilize CAD4COVID at both patient admissions, to support their diagnostic strategy and for the ICU to detect the temporal changes of the disease. In addition, we are planning to use CAD4COVID as a research tool. We seek to investigate the utility of AI support in COVID diagnostic workflow and the diagnostic accuracy of CAD4COVID in various settings."

– Pál Maurovich, Director of Medical Imaging Centre

In Indonesia, with a population of over 270 million people, the infectious nature of the corona virus is of particular concern. To combat the pandemic, there is a need for technological innovations that can help to make a difference.

The Faculty of Medicine at Universitas Indonesia, in collaboration with Dr. Cipto Mangunkusumo National General Hospital, have partnered with us to validate the CAD4COVID software.

Both organisations are also supporting the validation of CAD4TB, which has already been used to screen for tuberculosis across prisons in Jakarta; a project for which an MoU was signed last March between the Ministry of Health in Indonesia, Delft Imaging and its local partner IDBH Senso.

The Faculty of Medicine at Universitas Indonesia and Dr. Cipto Mangunkusumo National General Hospital have worked out a two-phased validation study to validate the performance of the CAD4COVID software, as well as its potential use case to combat COVID-19 in Indonesia.

The first phase of the validation study has recently been completed, after which the organisations have concluded: "The initial utilization of the AI toolbox at our hospital demonstrates that CAD4COVID has the potential to be implemented as a screening tool to detect suspected individuals of COVID-19 as it exhibits potential correlation with RT-PCR result. We recommend its usage in a resource-constrained environment."

Ken Singh from Glenwood Radiology regarding the use of CAD4COVID in their facility:

As of 19 May 2020, we have a total of 6716 cases with 1004 being active and 5584 being marked as recovered. 61 are in the hospital, out of which 8 are in intensive care. A total of 128 deaths have been recorded so far in Alberta due to COVID-19. With summer on the go and Alberta in its phase one of re-opening without any word on a functioning vaccine, we are anticipating a greater influx of patients in the coming months.

CAD4COVID is an excellent tool and we are very happy to be using it at our clinic. It directs to the areas of interest on the chest x-ray that can be further looked at in greater detail by our radiologists. We found it especially helpful for follow-up patients as we are able to see changes in their lung field pointed out through the software and the change in the CAD4COVID probability score. This is a good quantitative value that can be helpful in evaluating whether the patient’s condition is improving or worsening in addition to the image itself.

Dr. Euy Nyong Kin from the Korean Medical Foundation (KOMEF) in Seoul, has shared with us the following regarding our CAD4COVID solution:

"In Korea it is mandatory that people need to continue their daily lives and at the same time prevent the spread of COVID-19. We’ve been able to do this thanks to the support of CAD4COVID, a quick and easy to implement AI solution. Everywhere people could gather, not only in the hospital, quick and valid screening for COVID-19 should be imperative. I believe that at the entrance of every airport, hotel, train or restaurant people need to be checked by the simplicity of Artificial Intelligence by a quick and efficient chest X-ray. AI can play an extremely important role within the triaging phase of diagnostics. I support a world where people may have to submit a future ‘Health ID’ / ‘COVID ID’ that continuously needs to be updated on an annual basis. AI solutions such as CAD4COVID is a huge new happening for human survival and a first step into the right direction."

With now more than 6,000 confirmed cases across all Ghana , it was essential that Ghana would take measures to fight and contain the COVID-19 virus. Together with the support of the Netherlands Enterprise Agency (RVO) and Delft Imaging we will be installing the CAD4COVID software in 55 district hospitals across the country. This will further enhance its current healthcare infrastructure in order to face the pandemic.

Mieke Flierhuis (RVO): "In close cooperation with the Ministry of Health in Ghana, CAD4COVID will be installed in 55 district hospitals across the country. CAD4COVID analyzes images of the lungs and detects COVID-19 related abnormalities. The RVO, through its Dutch Good Growth Fund, supports this project."