What you need to understand about videofluoroscopy frame rates
Catriona M. Steele, Ph.D., S-LP(C), CCC-SLP, BCS-S, Reg. CASLPO
In 2007, I was responsible for writing the CASLPO Practice Standards and Guidelines (PSG) for Dysphagia. At the time, we were asked to include a recommendation regarding image acquisition rate in videofluoroscopy (also called “frame rate”), and we wrote the following:
“The temporal resolution (i.e., pulse rate) of videofluoroscopy should be determined in consultation with radiological personnel, balancing issues of radiation exposure with the need to capture a comprehensive dynamic recording of swallowing. Although current research evidence has not definitively identified the minimum temporal resolution necessary for imaging of the oropharyngeal swallow, it is suggested that pulse rates below 15 pulses per second may be insufficient to capture important events in swallowing. The video or digital recording of the dynamic swallowing study should be captured and archived at a minimum temporal resolution of 30 frames per second without compression so that adequate information regarding the swallow is available for later review.”
We summarized this guidance with the following practice guidelines:
In the 8 years that have passed since the publication of the PSG, questions regarding optimal image acquisition rate have been the focus of research conducted by Dr. Bonnie Martin-Harris, Dr. Heather Bonilha, Dr. Rosemary Martino, myself and others. Additionally, it is common for clinicians to seek guidance regarding frame rate. The purpose of this letter is to clarify several important points that have become clear as a result of research in the field.
1) Frame rate is NOT the same thing as fluoroscopy rate or pulse rate
When you are considering image acquisition rate for videofluoroscopy, it is important to know that several different terms may be used. Frame rate is a term that was historically used in a generic sense to describe the number of images that were generated per second in a videofluoroscopy. Advances in radiological imaging technology have introduced the need to be clear about the differences between fluoroscopy rate, pulse rate and frame rate.
Fluoroscopy rate refers to the number of images that are produced by the fluoroscope each second. On older, analog systems, the radiation beam that creates an x-ray image was either “on” or “off”, and when it was “on”, the radiation was continuous. The number of images that were produced per second depended on the frame rate of the recording system. In North America, standard video recording systems operate at a frame rate of 30 frames per second. Consequently, for historic (analog) assessments conducted using continuous fluoroscopy and a North American NTSC recording system, the output was 30 frames per second.
Digital fluoroscopy allows refinements to the way in which the radiation beam is delivered to the patient. Specifically, the radiation beam can be delivered as a sequence of short pulses. This has several potential benefits. First of all, when a dynamic recording is generated as a sequence of short pulsed images, the resulting recording is less susceptible to blur from motion artifact. Second, because there are actually gaps of no radiation between each radiation pulse, the overall radiation exposure will be reduced when pulsed fluoroscopy is used compared to a continuous fluoroscopy recording of the same length. This development has created an opportunity to limit radiation exposure by manipulating pulse rate to output fewer images per second. This is an important advantage given the guiding principle of keeping radiation exposure “As Low As Reasonably Achievable” (ALARA), but it raises the need to understand the pros and cons of lower pulse rates for dynamic radiological studies like oropharyngeal swallowing studies.
When a pulse rate of 30 pulses per second is used to generate 30 images per second, and these images are recorded at a frame rate of 30 frames per second, the visual results are the same as they would have been when continuous fluoroscopy was captured at a frame rate of 30 frames per second. That is, both recordings could be slowed down and played frame-by-frame, and in both scenarios you would have 30 unique images for every 1-second of recording. When you play either of these recordings at regular speed, your eye would be unable to discern the refresh rate as the recording proceeds from frame to frame. However, for two recordings of identical overall length, the version generated with a fluoroscopy pulse rate of 30 pulses per second would involve less radiation exposure to the patient than the continuous fluoroscopy version. The situation is actually, not quite that simple, and depends on additional parameters such as pulse width – but in principle, reducing radiation exposure by manipulating pulse rate is a possibility.
Modern fluoroscopy equipment comes with various options in image acquisition rates. It is possible for the radiologist or radiation technologist to select a pulse rate (typically 30, 15, 7.5, 4 or 2 pulses per second). It is also possible for the frame rate of the recording equipment to be set at 30, 15, 7.5, 4 or 2 frames per second. If you are working with equipment manufactured in Europe or Japan, you may find that there are image acquisition rates of 25 images per second (or fractions of this rate, e.g. 12, 6 and 3 images per second), based on the fact that video frequencies in these regions are not in NTSC format. If you are recording your videofluoroscopy directly to a hospital electronic archiving system (or PACS), you need to know what frame rate settings have been set for these recordings, which are frequently limited to fewer than 30 images per second with the goal of saving memory demands on the storage system. Dynamic radiological studies like videofluoroscopy take up much more storage space than still radiographic images. A further issue to be aware of is something called a “video loop”. A video loop stores a set number of frames in a working memory on the fluoroscopy equipment. As you approach the end of the loop, the operator needs to use a save function to transfer the contents of the loop to the recording archive. If the save function is not used, the video loop will slide forward, maintaining the set number of frames in working memory but dropping the earliest frames in the loop as it captures new frames on the opposite end.
These options set up some interesting opportunities and also some potential limitations for videofluoroscopy, which are important for you to consider as the clinician. The goal, of course, should always be to conduct a videofluoroscopy using settings that enable you to capture the necessary medical information to answer your questions, while being responsible about limiting unnecessary radiation exposure.
2) What is the optimum image acquisition rate?
In order to determine the best image acquisition rate for videofluoroscopy, we need to consider how short the shortest event of interest might be in swallowing, and what the consequences might be of missing that event. Recent publications have attempted to address this question by comparing the ability of speech-language pathologists to detect penetration-aspiration events on recordings with 15 and 30 images per second. This was cleverly done by recording videofluoroscopy at 30 frames per second, and then manipulating the recording to delete every 2nd image, thereby creating a 15 image per second version. There are three main findings from these studies:
a) Penetration-aspiration events are more frequently missed in recordings with only 15 images per second. This suggests that penetration-aspiration events can sometimes be extremely brief (i.e., shorter than 1/15 of a second).
b) For patients who have penetration-aspiration, fewer swallows are required at 30 images per second to catch the problem than at 15 images per second. This finding has been used to suggest that radiation exposure may, in fact, NOT end up being lower at 15 pulses per second, particularly if extra swallows are needed to reveal the problem.
c) Inter-rater agreement is reported to be higher when there are only 15 images per second. This is actually not an unexpected result, because there is less opportunity for disagreement across raters when fewer images are being reviewed. The finding is not dissimilar to the report that inter-rater agreement regarding penetration-aspiration is higher when binary decisions are made compared to the full 8-point Penetration-Aspiration Scale. It is important to remember that agreement and reliability across raters is something different than rating accuracy.
Although these research findings do not provide explicit guidance for clinicians who are trying to decide on the best image acquisition rates to use in their videofluoroscopies, they do raise some important questions for you to consider:
a) How serious do you think it would be if you missed a brief penetration-aspiration event because you were only capturing 15 (or fewer) images per second?
For some patients, you might decide that this is not actually very serious, because a very brief penetration-aspiration event is more likely to involve only a tiny amount of material entering the airway. If a penetration-aspiration event is not visible at 15 images per second, perhaps it is not significant enough to warrant intervention.
On the other hand, there may be some patients where respiratory concerns are very serious, and you want to be very confident that you have not missed any penetration-aspiration events. In these cases, 15 images per second are probably inadequate to rule out penetration-aspiration as a problem. Certainly, image acquisition rates below 15 images per second should be considered inadequate in terms of sensitivity for detecting penetration-aspiration.
b) How serious is the additional radiation exposure required to generate 30 images per second?
This question deals directly with the radiological ideal of “As Low As Reasonably Achievable”. In some cases, (for example if working with very young children), there may be valid reasons to limit radiation exposure by using pulse rates of 15 pulses per second. However, 15 images per second should be considered the lower limit, and dropping below this value is likely to miss important information. It is also important to remember that you might be able to answer your clinical questions more efficiently, i.e., in fewer swallows, if you pulse and record at 30 images per second, so the argument that radiation is saved at lower pulse rates may not be valid.
c) What else can be missed when fewer than 30 images per second are captured?
Swallowing is a dynamic physiological process, and one purpose of a videofluoroscopy is to look beyond the functional outcomes of penetration-aspiration or residue, to reveal the underlying mechanisms behind these problems. A second goal of the videofluoroscopy may be to explore the benefits of maneuvers for addressing pathophysiology. If fewer than 30 images per second are recorded, this will introduce constraints on your ability to discern the mechanisms behind impairment. Consider the following examples of potential limitations (note that this list is not comprehensive):
I. In healthy swallowing, a bolus typically travels the entire length of the pharynx (from the ramus of mandible down to the upper esophageal sphincter) in 1/30 of a second. Given that this is the case, recordings at lower image acquisition rates will not allow you to confidently determine timing measures like swallow response time (also called swallow delay, or stage transition duration) or to determine where the bolus head is located at the onset of the pharyngeal swallow.
II. Thicker liquids are thought to help swallowing safety by moving more slowly. If image acquisition rates of less than 30 images per second are used, you may not be able to detect differences in bolus flow rate between thin and thicker liquids.
III. A very critical question in understanding the pathophysiology behind penetration-aspiration is to determine when the problem occurs relative to closure of the laryngeal vestibule. At image acquisition rates lower than 30 images per second, you may not have enough information to determine whether the problem is one of late laryngeal vestibule closure, or to discern whether this problem is effectively addressed with thicker liquids.
Summary and Conclusions
I hope that this letter has helped to clarify some of the questions that currently exist regarding image acquisition rates in videofluoroscopy. Secondly, I hope this letter equips you to have informed conversations with your radiology colleagues when determining the settings to use in your facility. A list of references will follow at the end of the letter.
My own opinion is that the wording of the 2007 PSG continues to provide appropriate and helpful guidance in this regard. Image acquisition rates below 15 images per second are likely to be inadequate to capture important events in swallowing. Image acquisition rates of 30 images per second are likely to be optimal in terms of sensitivity to penetration-aspiration, and when generated using pulse rates of 30 pulses per second and frame rates of 30 frames per second are likely to capture impairment most efficiently, requiring the smallest number of swallows to reveal the problem.
Catriona M. Steele, Ph.D.
References that you may find helpful on this topic:
Bonilha, H. S., Blair, J., Carnes, B., Huda, W., Humphries, K., McGrattan, K., Martin-Harris, B. (2013). Preliminary investigation of the effect of pulse rate on judgments of swallowing impairment and treatment recommendations. Dysphagia, 28(4), 528-538.
Bonilha, H. S., Humphries, K., Blair, J., Hill, E. G., McGrattan, K., Carnes, B., . . .Martin-Harris, B. (2013). Radiation exposure time during MBSS: Influence of swallowing impairment severity, medical diagnosis, clinician experience, and standardized protocol use. Dysphagia, 28(1), 77-85.
Bushberg, J. T., Seibert, J.A., Leidholdt, E. M. Jr. & Boone, J. M. (2012). The Essential Physics of Medical Imaging. 3rd Edition. Lippincott Williams & Wilkin. Philadelphia, PA.
CASLPO. (2007). Practice Standards and Guidelines for Dysphagia Intervention by Speech- Language Pathologist. http://www.caslpo.com/sites/default/uploads/files/PSG_EN_Dysphagia.pdf
Cohen, M. (2009). Can we use pulsed fluoroscopy to decrease the radiation dose during video fluoroscopy feeding studies in children? Clinical Radiology, 64, 70-73.
Jaffer, N., Au, F. W., Ng, E. & Steele, C. M. (2015). Fluoroscopic evaluation of oropharyngeal dysphagia. AJR American Journal of Roentgenology, 204, 49-58.
Martino, R., Shaw, S., Greco, E., Maki, E., Jabbour, N., Gomes, A.,… Ringash, J. (2015). Comparing physiological swallow measures captured on videofluoroscopy at different frame rates: A reliability analysis. Oral presentation. 22nd Dysphagia Research Society meeting. Chicago, IL.
Peladeau-Pigeon, M. & Steele, C. M. (2013). Technical aspects of a videofluoroscopic swallowing study. Canadian Journal of Speech-Language Pathology and Audiology, 37(3), 216-226.
Steele, C. M., Peladeau-Pigeon, M. & Tam, K. (2015). When 30 frames per second is not 30 images per second. Poster presentation. 22nd Dysphagia Research Society meeting. Chicago, IL.
Peladeau-Pigeon, M. & Steele, C. M. (June, 2015, in press). Understanding Image Resolution and Quality in Videofluoroscopy. Perspectives (ASHA Special Interest Group 13).