A Clinical Validation Study of a Computerized Movement Analysis of the Colonoscope

Objectives:

1. Gather validity evidence for the 3D-Colonoscopy Progression score (3D-CoPS) and 3D-Colonoscopy Retraction Score (3D-CoRS) in a clinical setting.
2. Investigate the relationship between 3D-CoPS/3D-CoRS and a retrospective calculated Adenoma Detection Rate (ADR) of endoscopists with various experience.

Background:

Colonoscopy is the gold standard for diagnosing adenomas and colorectal cancer. High Adenoma Detection Rate (ADR) is essential to reduce the risk of subsequent colorectal cancer .To maximize the detection of adenomas during a colonoscopy the endoscopist need to see as much mucosa as possible. To increase the mucosa visualized quality indicators, such as bowel preparatin, cecal intubation rate and an average time of at least 6 minutes spent during withdrawal, should be used as supportive measures to minimize missed lesions and thereby decreasing interval cancer. Bowel preparation, cecal intubation and time spent during withdrawal are prerequisite measures for optimal visualization of the mucosa in the colorectum, but are not a quality measure for actual detected pathology. Time as a quality assurance for detection of adenomas means nothing if the physician isn't careful during withdrawal and look behind folds and change position to get a clear view of the mucosa surface. The ADR has been introduced as a surrogate measure of adequate visualization of the mucosa, but a recommend and a minimum number of 500 colonoscopies are needed to calculate a statistic reliable ADR. Although ADR is one of the most widely used and accepted quality indicators great variance in number of performed procedures and ADR among physicians exits, which makes ADR difficult to use in practice. Currently there is no quality parameter evaluating the individual procedure and to our knowledge no objective tools exits to help increase diagnostic accuracy during a colonoscopy. Objective, easy implementable and automatic measures are needed to ensure a high detection of adenomas and subsequently prevent missed lesions and interval cancers.

In collaboration with the Technical University of Denmark we have developed a score of progression and retraction during a colonoscopy, called 3D-Colonoscopy Progression Score (3D-CoPS) and 3D-Colonoscopy Retraction Score (3D-CoRS). We predict the systems to be automated and objective tool correlated with the endoscopists technical level of expertise in clinical colonoscopy.

Method and materials:

Retrospective study part:

The study is conducted at three different university hospitals. Twenty endoscopists with a lifetime experience of no less than 50 screening colonoscopies are voluntarily enrolled. To calculate the ADR for each endoscopist data of the last screenings colonoscopies is retrospectively collected (minimum 50 colonoscopies per endoscopist).

The following will be noted for each endoscopist:

• Number of total colonoscopies
• Number of colonoscopies within the last year
• Number of other endoscopies
• Number of screenings colonoscopies
• Intern, junior resident, senior resident, consultant
• Surgical or medical gastrointestinal background
• Sex
• Age
• Time since graduation

Prospective part:

A minimum of five consecutive screenings colonoscopies for each endoscopist are included. Patients appointed to a screening colonoscopy are chosen due to a higher ADR (approximately 50%) and a more homogeneous group. All procedures are performed with a standard Olympus colonoscope model in combination with the Magnetic Endoscopy Imaging system (MEI). Data collection starts at the intubation of the colonoscope in the anus and stops when the procedure is finished. The data collection is fully automated without any interference from the investigators during the normal clinical setup.

Approval is applied at the Danish Data Protection Agency, the Danish Patient Safety Authority and the Regional Committee on Health Research Ethics.

General mathematical formulation for 3D-CoPS and 3D-CoRS:

Electromagnetic coils are built in along the length of the Olympus colonoscopes. They generate a pulsed low-intensity magnetic field that is picked up a receiver coil. The data-points for each coil are inserted into an algorithm for movement analyzing. The two main measurements conducted in relation with the colonoscopy is the Colonoscopy Progression Score (CoPS) and Colonoscopy Retraction Score (CoRS). Both of these have a joint basis for the analysis, in the form of a movement analysis. This analysis is done as a vector product between the tip of the scope, and the next tracked magnetic coil - where on the relative magnitude of the change in position is multiplied on. This results in a relative movement that is scaled with how much the scope moves in relation to the previous position, and how much this movement is deflected according to the direction the tip points in. This movement analysis is used for distinction between the progression- and retraction phase, which is essential, such that the CoPS and CoRS measurements is performed at the right times.

3D-CoPS: To perform the CoPS measurement, the above definition helps by defining the progression phase. The general belief is that if the scope progresses smoothly, the colonoscopist is experienced, and vice versa. Therefor tracking how much time the tip of the scope is "stagnant", is used to define the CoPS. Specifically, a two dimensional histogram is used, spanned across the movement area of the colonoscopy. Filling the histogram is simply performed by sampling the tip position at each time step, and putting it into the respective bin. If the scope is then stagnant at many stages throughout the procedure, the bins values will be locally very high. On the other hand, if a smooth progression is present, the bin values will be smoothly distributed, across all the different bins. This is the basis for the CoPS value. Finally followed with a normalizing, and scaling of the 2D histogram value, for easing interpretation.

3D-CoRS: Analyzing the retraction phase, must be done differently, as the tip is supposed to be stagnant, at different times, through this phase. Therefor the movement analysis, as described above, is used to measure how the instrument moves. This movement is then high-passed filtered, as experienced colonoscopist seems to have much more high frequency movements, where-in the novices moves with much lower frequency (more hesitant). This overall movement is evaluated as a difference between the different positions (by differentiation), and a count of peaks and valleys is done with a fitting margin. This final count is then high if an expert is performing the procedure, as more high frequency movements will take place, and low if a novice is performing the procedure.

Patient related information:

• Number of polyps
• The polyp size
• Polyp type
• Polyp histology
• Polyp location

Validity:

Evidence of validity is based on Messicks five major sources (10).

Content:

1. 3D-CoPS and 3D-CoRS are being developed based on the evidence of an ongoing simulation-based validation study " A Validation Study of a Computerized Movement Analysis of the Colonoscope in Simulated Colonoscopy" and general good practice of how a colonoscope is withdrawn from cecum to anus.

Response process:

The process is standardized using an Olympus colonoscope and MEI system. All participants will be instructed informed of the systems and the data-collection. All data-collection regarding are presented in a uniform file-format. The 3D-CoPS and 3D-CoRS are objective, automated and unbiased tools.

Internal structure:

To ensure internal consistency of the 3D-CoPS and 3D-CoRS among the endoscopists a minimum of 5 consecutive colonoscopies are needed (based on an article not yet published). The score system is automated and unbiased why the internal structure in this regard is consistent.

Relationship to other variables:

We assume the sample data reflects a population that follows a probability distribution based on a fixed set of parameters. The correlation between experience (logarithmic) and scores will be investigated using Pearson's r. Scores from 3D-CoPS and 3D-CoRS will be correlated with patient and endoscopist variables. The endoscopist is not blinded during data-collection, which might affect the performance. Even thou data collection is fully automated, the technic during withdrawal of the colonoscope from caecum to anus could change and increase mucosa visualized and therefore polyp detection (increased awareness of the endoscopist).

Consequences:

Contrasting groups' method will be used to set a pass/fail standard based on the scores imitating aminimum performance during intubation, a minimum ADR and false positive and false negative will be investigated.

Ethics:

The study contains no biomedical involvement. Participants and patients will not suffer from any physical or psychological discomfort. All endoscopists will be provided with oral and written information and sign a letter of informed consent before entering the study. The collected data are anonymized.