Osteoporosis is under-screened and under-diagnosed. With Rho, we aim to enable the initiation of preventative care earlier in the course of the disease and ultimately reduce fracture incidence. Rho is a medical device that identifies patients who would benefit from clinical fracture risk assessment from routinely acquired x-ray imaging. These are x-ray images that were acquired for other reasons (i.e chest x-ray to rule out pneumonia). In this way, Rho operates as an opportunistic screening system by generating new value from existing data.
Rho leverages modern machine learning to identify patients who would benefit from clinical fracture risk assessment. To train the model, researchers utilized over 60,000 x-ray and DXA pairs performed within 1 year of each other in outpatients at multiple imaging facilities over a 10 year period.
Rho was validated in an independent test set of over 2000 patients. The area under the receiver operating curve (AUC) was used as the primary endpoint.
Rho is a Health Canada-licensed software medical device that analyzes a standard x-ray using artificial intelligence to screen patients who are likely to have low bone mineral density (BMD). The output of Rho is a report that is provided to a radiologist when they are reviewing an x-ray. The report presents a Rho Score between 1 and 10. A higher Rho Score indicates a higher likelihood of the patient having a DXA-derived T-score < -1 in at least one of the femoral neck or lumbar spine (L1-L4). This is meant to aid health care professionals in the assessment of risk of low bone mineral density (BMD). Rho produces adjunctive information. It is not a diagnostic aid.
Rho is an opportunistic screening tool that analyzes frontal radiographs of the lumbar spine, thoracic spine, chest, pelvis, knee, or hand performed for any clinical indication in patients aged 50 years and older. The algorithm was developed using machine learning to generate a risk score that correlates with a patient's likelihood of having low bone mineral density.
If your patient is aged 50 years and older and had an x-ray of the chest, pelvis, knee, lumbar spine, thoracic spine, or hand at an institution that has adopted Rho, it will automatically be analyzed as soon as the x-ray is performed. The reporting radiologist reviews this information and can include this in the x-ray report.
As a radiologist, you can choose to include this finding in your report of the x-ray.
As a care provider receiving this information, you can consider conducting a clinical fracture risk assessment. Based on this assessment, you can decide whether or not the patient would benefit from a DXA to accurately quantify the patient’s BMD and fracture risk.
Osteoporosis is a silent disease resulting in fractures which occur in 1 in 3 women and 1 in 5 men (https://osteoporosis.ca/what-is-osteoporosis/). Screening for osteoporosis is under utilized; 80% of patients who sustain a fragility fracture have never been screened before the incident (Gillespie CW, Morin PE. Am J Med. 2017). Guidelines for screening with DXA vary by jurisdiction due to cost, and preventative care is often overlooked in today’s healthcare environment. As most patients over age 50 undergo an x-ray for some reason, Rho can act as an opportunistic pre-screening tool utilizing this already captured data to identify patients who are more likely to benefit from further screening. With Rho, we aim to enable the initiation of preventative care earlier in the course of the disease and ultimately reduce fracture incidence.
Radiologists are interested in Rho because it allows them to improve patient care, provide a better service to their referral base, and increase reimbursable services. All of this is done with minimal impact to reporting workflow.
Rho automatically identifies when an eligible x-ray has been performed at an institution, analyzes it, and sends a Rho Report to the accession if the patient is likely to have low BMD. The Rho Report is directly viewable in the PACS. When reporting the x-ray, the radiologist can review the Rho Report, and, if in agreement, can simply include this finding in their own words, or insert a pre-programmed macro through their voice dictation software. To see Rho in action, please review this 1 minute video. The Rho Score threshold that triggers reports to be sent to PACS is configurable by the institution.
With the default configuration, the Rho report will often be viewable in PACS within 4 minutes of the x-ray being uploaded to PACS. This latency can be customized based on your site’s requirements and IT infrastructure.
Rho has a configurable parameter which will prevent analysis of a patient who has had a DXA within X number of months. The default value for this is 24 months.
Rho will automatically identify this and will only send a single Rho Report for that accession.
In that case, the patient may receive multiple Rho Reports to different accessions. There will never be more than 1 report per accession. Rho has an AUC of 0.8 or more for each body part, but there will be times when a Rho Score for the same patient from x-rays of two different body parts will not match. Rho is not designed to diagnose or rule out disease. We believe that any suggestion of increased risk of low BMD warrants reporting, as the consequences are a simple clinical fracture risk assessment, however the radiologist is free to use their discretion.Optionally, Rho can be configured to send one report per patient within a specified timeframe (e.g., per day or per week).
A typical radiologist can only detect osteopenia on conventional radiographs when 20-40% of bone mass has been lost. If someone had a BMD of 0.858 g/cm2 at the femoral neck (which is the mean BMD for a female aged 20-29 years from NHANES III), then a loss of 30% of bone mass would correspond to a T-Score of -2.1 (assuming the use of a female reference population as recommended by WHO). Similarly, if someone had a BMD of 1.064 g/cm2 at L1-L4 (which is the mean BMD for a female aged 20-29 years from NHANES), then a loss of 30% would correspond to a T-Score of -3.0. As such, an algorithm that can identify the earlier stage of demineralization, at a T-score of -1, offers an advantage to the human eye.
A typical radiologist can only detect osteopenia on conventional radiographs when 20-40% of bone mass has been lost. If someone had a BMD of 0.858 g/cm2 at the femoral neck (which is the mean BMD for a female aged 20-29 years from NHANES III), then a loss of 30% of bone mass would correspond to a T-Score of -2.1 (assuming the use of a female reference population as recommended by WHO). Similarly, if someone had a BMD of 1.064 g/cm2 at L1-L4 (which is the mean BMD for a female aged 20-29 years from NHANES), then a loss of 30% would correspond to a T-Score of -3.0. As such, an algorithm that can identify the earlier stage of demineralization, at a T-score of -1, offers an advantage to the human eye.
Current recommendations are that patients 50 years and over undergo clinical fracture risk assessments and patients 65 and over be considered for DXA, but in practice, preventative care is not always top of mind. Flagging all adult x-ray patients in need of clinical fracture risk assessment is no different than the existing guidelines which are well-known by our clinical colleagues. Rho helps radiologists identify the patients who are most likely to benefit from this risk assessment which helps clinicians prioritize their limited time. The table below shows that when age is used as a predictor of low BMD, the area under the receiver operating characteristic curve (AUC) is about 0.6 in both females and males, but using the Rho Score as the predictor of low BMD, the AUC is significantly higher.
The frequency of low BMD at each Rho Score was derived from a retrospective, multi-center trial to assess its performance at assessing a patient’s likelihood of having low BMD from frontal radiographs, using DXA as the benchmark with which low BMD is diagnosed (at least one of the femoral neck or L1-4 having a T-score <-1). Assessments were made with a set of x-ray/DXA pairs that included 6 types of x-ray that Rho can analyze (lumbar spine, thoracic spine, chest, pelvis, knee, or hand) in males and females. Performance was assessed in dataset hat independent from the dataset used to train Rho. The dataset included participants with a range of BMD values that cover the entire range of disease state, but included more participants with low BMD than the general population.
The sample sizes of TNI and OMN (3758 and 513 x-ray/DXA pairs, respectively; one per patient), had over 80% power at a one-sided alpha of 0.025 to detect an area under the receiver operating curve (AUC) point estimate of 0.85 and a performance goal of 0.75.
AUC and 95% confidence intervals are presented for the entire TNI dataset, females and males, and from x-rays of the six body parts that Rho can analyze.
Our current pricing model is value-based, where you only pay for Rho when patients return to your institution for DXA within a certain time period of an x-ray examination which has had a Rho report sent under the same accession (i.e. they are “Rho positive”).
Invoices for Rho are generated at the beginning of the month for the prior month, and will be sent by email to the specified contact on file.
Payments are due within 30 days. Payments can be made through ACH, ETF or credit card. We are currently offering Rho with an Early Adopter promotion. This promotion waives installation, servicing, and all other upfront fees.
Once a purchasing decision has been made we will:
- Send you our standard contract, often within 1 business day.
- Coordinate with your IT administrator to ensure the modest technical requirements are met (see IT Administrators section for more details).
- Conduct Rho installation, IT Admin training, and configuration (often completed in 1 hour).
- Circulate brief training material for Radiologists.
Motivated sites can be up and running within 1 week of a purchase decision.
It is technically possible to run Rho on patients who have undergone x-rays retrospectively. A list of patients who have not undergone DXA in a specific period of time and ranked by Rho score can be provided to you to perform outreach or provide these to your referral base. Please contact us to discuss specific details.
Rho can be installed on-premise or in the cloud, however, it must be within your existing secured network. All traffic to and from Rho can be monitored and controlled by you. Rho consists of a DICOM node, database and a number of microservices deployed through container orchestration services like Docker Compose and Kubernetes. Most installations will configure any DICOMS sent to Rho to be purged daily. Generally, Rho poses no additional security risks beyond those that exist for all services operating within modern health IT networks. With de novo granting from the FDA as a SaMD, Rho has undergone a cybersecurity risk assessment. Contact 16 Bit for a list of cybersecurity features built into Rho, and cybersecurity recommendations for customers
Rho requires a simple physical or virtual machine with the following minimum specifications:
- 64 bit 2 GHz dual core processor CPU with AVX support
- 16 GiB RAM (system memory)
- 250 GB of hard drive space
- Linux-based operating system that supports Kubernetes (e.g., Ubuntu 22.04), or
Windows 10 with Docker Desktop
- Internet connection
- VPN access preferred for servicing
- Optionally: NVIDIA cuda-enabled GPU
Installation, IT Administrator training, and configuration is often performed within 1 hour.
Rho is configured to perform automatic daily backups, weekly backups, and cleaning of the hard drive. Rho notifies 16 Bit when these tasks are performed and if they are successful. As such, there are no regular servicing requirements. Software updates can be installed remotely via VPN.
IT administrators are trained during the installation and configuration of Rho. This is done using a 20 minute interactive presentation. A similar presentation is provided to the customer to circulate to the radiologists that will be interacting with Rho. 16 Bit can provide additional training sessions upon request.
IT Administrators will not have to interact with Rho often. At time of installation, 16 Bit helps you configure Rho for your institution. But very occasionally, your team might want to change a setting, which you can do through its intuitive user interface. 16 Bit can also help with this, if needed. Radiologists do not interact with the Rho user interface, they only see Rho Reports through PACS.
An IT Administrator, or other interested party, can access the Rho Dashboard through a web browser on the same network as Rho. This user interface displays a breakdown of which x-rays have been analyzed, whether a report was sent, and when a Rho-screened patient returns for DXA.
