1
|
Yi X, Adams S, Babyn P, Elnajmi A. Automatic Catheter and Tube Detection in Pediatric X-ray Images Using a Scale-Recurrent Network and Synthetic Data. J Digit Imaging 2020; 33:181-190. [PMID: 30972586 PMCID: PMC7064683 DOI: 10.1007/s10278-019-00201-7] [Citation(s) in RCA: 12] [Impact Index Per Article: 3.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/27/2022] Open
Abstract
Catheters are commonly inserted life supporting devices. Because serious complications can arise from malpositioned catheters, X-ray images are used to assess the position of a catheter immediately after placement. Previous computer vision approaches to detect catheters on X-ray images were either rule-based or only capable of processing a limited number or type of catheters projecting over the chest. With the resurgence of deep learning, supervised training approaches are beginning to show promising results. However, dense annotation maps are required, and the work of a human annotator is difficult to scale. In this work, we propose an automatic approach for detection of catheters and tubes on pediatric X-ray images. We propose a simple way of synthesizing catheters on X-ray images to generate a training dataset by exploiting the fact that catheters are essentially tubular structures with various cross sectional profiles. Further, we develop a UNet-style segmentation network with a recurrent module that can process inputs at multiple scales and iteratively refine the detection result. By training on adult chest X-rays, the proposed network exhibits promising detection results on pediatric chest/abdomen X-rays in terms of both precision and recall, with Fβ = 0.8. The approach described in this work may contribute to the development of clinical systems to detect and assess the placement of catheters on X-ray images. This may provide a solution to triage and prioritize X-ray images with potentially malpositioned catheters for a radiologist's urgent review and help automate radiology reporting.
Collapse
Affiliation(s)
- X Yi
- College of Medicine, University of Saskatchewan, Saskatoon, SK, Canada.
| | - Scott Adams
- College of Medicine, University of Saskatchewan, Saskatoon, SK, Canada
| | - Paul Babyn
- College of Medicine, University of Saskatchewan, Saskatoon, SK, Canada
| | - Abdul Elnajmi
- College of Medicine, University of Saskatchewan, Saskatoon, SK, Canada
| |
Collapse
|
2
|
Impact of Patient Photos on Detection Accuracy, Decision Confidence and Eye-Tracking Parameters in Chest and Abdomen Images with Tubes and Lines. J Digit Imaging 2019; 32:827-831. [PMID: 30891651 DOI: 10.1007/s10278-019-00202-6] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 10/27/2022] Open
Abstract
To minimize errors in imaging studies, a camera system was developed that acquires images of patients simultaneously with radiographic images. Thirty-seven chest/abdomen portable radiographs showing central lines, orogastric/nasogastric/endotracheal tubes with patient photographs were viewed by six radiologists while eye-position was recorded. They indicated whether each line/tube was present/absent and rated confidence. Images were shown in three conditions: radiograph only, small, or large photograph with radiograph. There was greater accuracy in detecting tubes with photographs present and decision confidence was generally higher with the photographs. For the eye-tracking parameters, total viewing time, number of fixations, and number of times observers transferred viewing from radiograph to photograph differed as function of whether a photograph was present or absent as well as photograph size. Adding patient photographs to radiographic interpretation of chest and abdomen films can aid in the detection of tubes/lines. If photograph size is large enough, it takes an average of only 3 extra seconds to view compared to the radiograph alone and adds significant confidence to decisions.
Collapse
|
3
|
Lee H, Mansouri M, Tajmir S, Lev MH, Do S. A Deep-Learning System for Fully-Automated Peripherally Inserted Central Catheter (PICC) Tip Detection. J Digit Imaging 2018; 31:393-402. [PMID: 28983851 PMCID: PMC6113157 DOI: 10.1007/s10278-017-0025-z] [Citation(s) in RCA: 26] [Impact Index Per Article: 4.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/29/2022] Open
Abstract
A peripherally inserted central catheter (PICC) is a thin catheter that is inserted via arm veins and threaded near the heart, providing intravenous access. The final catheter tip position is always confirmed on a chest radiograph (CXR) immediately after insertion since malpositioned PICCs can cause potentially life-threatening complications. Although radiologists interpret PICC tip location with high accuracy, delays in interpretation can be significant. In this study, we proposed a fully-automated, deep-learning system with a cascading segmentation AI system containing two fully convolutional neural networks for detecting a PICC line and its tip location. A preprocessing module performed image quality and dimension normalization, and a post-processing module found the PICC tip accurately by pruning false positives. Our best model, trained on 400 training cases and selectively tuned on 50 validation cases, obtained absolute distances from ground truth with a mean of 3.10 mm, a standard deviation of 2.03 mm, and a root mean squares error (RMSE) of 3.71 mm on 150 held-out test cases. This system could help speed confirmation of PICC position and further be generalized to include other types of vascular access and therapeutic support devices.
Collapse
Affiliation(s)
- Hyunkwang Lee
- Department of Radiology, Massachusetts General Hospital, 25 New Chardon Street, Suite 400B, Boston, MA 02114 USA
| | - Mohammad Mansouri
- Department of Radiology, Massachusetts General Hospital, 25 New Chardon Street, Suite 400B, Boston, MA 02114 USA
| | - Shahein Tajmir
- Department of Radiology, Massachusetts General Hospital, 25 New Chardon Street, Suite 400B, Boston, MA 02114 USA
| | - Michael H. Lev
- Department of Radiology, Massachusetts General Hospital, 25 New Chardon Street, Suite 400B, Boston, MA 02114 USA
| | - Synho Do
- Department of Radiology, Massachusetts General Hospital, 25 New Chardon Street, Suite 400B, Boston, MA 02114 USA
| |
Collapse
|
4
|
Abstract
Perceptual errors are common contributors to missed diagnoses in the clinical practice of radiology. While the physical attributes of an image such as image resolution, signal-to-noise characteristics, and anatomic complexity are major causes of poor conspicuity of pathologic lesions, there are major interrelated cognitive contributors to visual errors. The first is satisfaction of search (SOS), where the detection of an abnormality results in premature termination of further search. Another form of incomplete search pattern is visual isolation, where a radiologist's search pattern is truncated to the main areas of an image, while little or no attention is given to peripheral areas. A second cognitive error is inattentional blindness, defined as the failure to notice a fully visible, but unexpected object because attention was otherwise engaged. Strategies for error mitigation have centered around the use of check lists, self prompting routines, and structured reporting within an institutional culture of safety and vigilance.
Collapse
Affiliation(s)
- George A Taylor
- Department of Radiology, Harvard Medical School, Boston Children's Hospital, 300 Longwood Avenue, Boston, MA 02115, USA
| |
Collapse
|
5
|
Pediatric radiology malpractice claims - characteristics and comparison to adult radiology claims. Pediatr Radiol 2017; 47:808-816. [PMID: 28536766 DOI: 10.1007/s00247-017-3873-2] [Citation(s) in RCA: 7] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 02/06/2017] [Revised: 03/16/2017] [Accepted: 04/19/2017] [Indexed: 10/19/2022]
Abstract
Medical malpractice is the primary method by which people who believe they have suffered an injury in the course of medical care seek compensation in the United States and Canada. An increasing body of research demonstrates that failure to correctly diagnose is the most common allegation made in malpractice claims against radiologists. Since the 1994 survey by the Society of Chairmen of Radiology in Children's Hospitals (SCORCH), no other published studies have specifically examined the frequency or clinical context of malpractice claims against pediatric radiologists or arising from pediatric imaging interpretation. We hypothesize that the frequency, character and outcome of malpractice claims made against pediatric radiologists differ from those seen in general radiology practice. We searched the Controlled Risk Insurance Co. (CRICO) Strategies' Comparative Benchmarking System (CBS), a private repository of approximately 350,000 open and closed medical malpractice claims in the United States, for claims related to pediatric radiology. We further queried these cases for the major allegation, the clinical environment in which the claim arose, the clinical severity of the alleged injury, indemnity paid (if payment was made), primary imaging modality involved (if applicable) and primary International Classification of Diseases, 9th revision (ICD-9) diagnosis underlying the claim. There were a total of 27,056 fully coded claims of medical malpractice in the CBS database in the 5-year period between Jan. 1, 2010, and Dec. 31, 2014. Of these, 1,472 cases (5.4%) involved patients younger than 18 years. Radiology was the primary service responsible for 71/1,472 (4.8%) pediatric cases. There were statistically significant differences in average payout for pediatric radiology claims ($314,671) compared to adult radiology claims ($174,033). The allegations were primarily diagnosis-related in 70% of pediatric radiology claims. The most common imaging modality implicated in pediatric radiology claims was radiography. The highest payouts in pediatric radiology pertained to missed congenital and developmental anomalies (average $1,222,932) such as developmental dysplasia of the hip and congenital central nervous system anomalies. More than half of pediatric radiology claims arose in the ambulatory setting. Pediatric radiology is not immune from claims of medical malpractice and these claims result in high monetary payouts, particularly for missed diagnoses of congenital and developmental anomalies. Our data suggest that efforts to reduce diagnostic error in the outpatient radiology setting, in the interpretation of radiographs, and in the improved diagnosis of fractures and congenital and developmental anomalies would be of particular benefit to the pediatric radiology community.
Collapse
|
6
|
Taylor EC, Taylor GA. Radiographic signs of non-venous placement of intended central venous catheters in children. Pediatr Radiol 2016; 46:210-8. [PMID: 26637319 DOI: 10.1007/s00247-015-3462-1] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.1] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 05/15/2015] [Revised: 07/29/2015] [Accepted: 08/28/2015] [Indexed: 10/22/2022]
Abstract
BACKGROUND Central venous catheters (CVCs) are commonly used in children, and inadvertent arterial or extravascular cannulation is rare but has potentially serious complications. OBJECTIVE To identify the radiographic signs of arterial placement of CVCs. MATERIALS AND METHODS We retrospectively reviewed seven cases of arterially malpositioned CVCs on chest radiograph. These cases were identified through departmental quality-assurance mechanisms and external consultation. Comparison of arterial cases was made with 127 age-matched chest radiographs with CVCs in normal, expected venous location. On each anteroposterior (AP) radiograph we measured the distance of the catheter tip from the right lateral border of the thoracic spine, and the angle of the vertical portion of the catheter relative to the midline. On each lateral radiograph we measured the angle of the vertical portion of each catheter relative to the anterior border of the thoracic spine. When bilateral subclavian catheters were present, the catheter tips were described as crossed, overlapping or uncrossed. RESULTS On AP radiographs, arterially placed CVCs were more curved to the left, with catheter tip positions located farther to the left of midline than normal venous CVCs. When bilateral, properly placed venous catheters were present, all catheters crossed at the level of the superior vena cava (SVC). When one of the bilateral catheters was in arterial position, neither of the catheters crossed or the inter-catheter crossover distance was exaggerated. On lateral radiographs, there was a marked anterior angulation of the vertical portion of the catheter (mean angle 37 ± 15° standard deviation [SD] in arterial catheters versus 5.9 ± 8.3° SD in normally placed venous catheters). CONCLUSION Useful radiographic signs suggestive of unintentional arterial misplacement of vascular catheters include leftward curvature of the vertical portion of the catheter, left-side catheter tip position, lack of catheter crossover on the frontal radiograph, as well as exaggerated anterior angulation of the catheter on the lateral chest radiograph.
Collapse
Affiliation(s)
- Erin C Taylor
- Department of Radiology, Boston Children's Hospital, 300 Longwood Ave., Boston, MA, 02115, USA
| | - George A Taylor
- Department of Radiology, Boston Children's Hospital, 300 Longwood Ave., Boston, MA, 02115, USA.
- Department of Radiology, Harvard Medical School, Boston, MA, USA.
| |
Collapse
|
7
|
Chang PT, Taylor GA. Umbilical venous catheter malposition and errors in interpretation in newborns with Bochdalek hernia. Pediatr Radiol 2015; 45:982-8. [PMID: 25875634 DOI: 10.1007/s00247-014-3275-7] [Citation(s) in RCA: 8] [Impact Index Per Article: 0.9] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 08/21/2014] [Revised: 11/12/2014] [Accepted: 12/29/2014] [Indexed: 10/24/2022]
Abstract
BACKGROUND Neonates with congenital diaphragmatic hernia (CDH) often require placement of lines and tubes for supportive therapy. The resulting altered anatomy can result in diagnostic errors when interpreting the location of support lines and tubes such as UVCs (umbilical venous catheters). OBJECTIVE The purpose of this study was to evaluate the effect of CDH on UVC position and to evaluate the accuracy at which radiologists describe the position on chest radiographs. MATERIALS AND METHODS During a 5-year period, 406 chest radiographs performed within 7 days of birth in infants with congenital diaphragmatic hernia were identified and reviewed for the following data: presence of UVC, location of catheter tip (cavoatrial junction, intracardiac, intrahepatic or umbilical vein), and location of CDH (right or left). The radiologic report of the UVC tip location for each case was then reviewed individually to determine the adequacy of interpretation. Inadequate reports were classified as incorrect (the wrong location of the catheter tip was reported), no mention (the location of the catheter tip was in a suboptimal location but not mentioned), and not specified (the precise location of the catheter tip was not clearly stated in the report when the tip was in a suboptimal location). RESULTS A total of 60 infants were identified as having CDH (56 on the left, 4 on the right). The most common location for an incorrectly placed UVC was the contralateral chest, accounting for 26.7% (16/60) of the infants, followed by an abdominal intrahepatic location (16.7%) and the umbilical vein (8.3%). Thirty percent (120/406) of the chest radiograph reports were found to be inadequate regarding the interpretation of the location of the catheter tip. The majority of the inadequate reports (48/406, 11.8%) did not specify when the catheter tip was in a suboptimal location. In 37 reports (9.1%), the location of the catheter tip was reported incorrectly, and no mention of the catheter location was made in 35 reports (8.6%). CONCLUSION The location of an UVC in an infant with Bochdalek hernia can pose a diagnostic challenge because of the altered anatomy and change in the expected course of the catheter. Familiarity with the altered anatomy and vigilance of the various abnormal locations in which UVCs can be placed can help optimize management for the child and reduce morbidity and mortality.
Collapse
Affiliation(s)
- Patricia T Chang
- Department of Radiology, Boston Children's Hospital and Harvard Medical School, 300 Longwood Ave., Boston, MA, 02115, USA,
| | | |
Collapse
|
8
|
Engelkemier DR, Taylor GA. Pitfalls in pediatric radiology. Pediatr Radiol 2015; 45:915-23. [PMID: 25520014 DOI: 10.1007/s00247-014-3196-5] [Citation(s) in RCA: 4] [Impact Index Per Article: 0.4] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 05/21/2014] [Revised: 09/20/2014] [Accepted: 09/25/2014] [Indexed: 10/24/2022]
Abstract
This essay depicts some of the diagnostic errors identified in a large academic pediatric imaging department during a 13-year period. Our aim is to illustrate potential situations in which errors are more likely to occur and more likely to cause harm, and to share our difficult cases so other radiologists might learn without having to experience those situations themselves.
Collapse
Affiliation(s)
- Dawn R Engelkemier
- Department of Radiology, Boston Children's Hospital, Harvard Medical School, 300 Longwood Ave., Boston, MA, 02115, USA,
| | | |
Collapse
|
9
|
A review of published case reports of inadvertent pulmonary placement of nasogastric tubes in children. J Pediatr Nurs 2014; 29:e7-12. [PMID: 24071621 DOI: 10.1016/j.pedn.2013.08.009] [Citation(s) in RCA: 13] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 07/24/2013] [Accepted: 08/24/2013] [Indexed: 01/12/2023]
Abstract
PROBLEM Little is known about the incidence of inadvertent pulmonary placement of nasogastric tubes during blind insertions in children. PURPOSE The purpose of this paper was to conduct a review of published case reports over the past two decades. METHODS An OVID Medline search was conducted of articles published from 1993 through 2012. RESULTS Fifteen published case reports were located; four patients died as a result of their malpositioned tubes. The auscultatory bedside method failed to detect the malpositioned tubes in all seven cases in which it was used. CONCLUSIONS The incidence of inadvertent pulmonary placement of nasogastric tubes is relatively low but can lead to serious and even lethal results in children. The auscultatory method to predict tube location is unreliable.
Collapse
|
10
|
Gauguet JM, Prabhu SP, Pigula F, Voss SD. The radiographic appearance of split Blake drains: what you see is not necessarily what you get. Pediatr Radiol 2013; 43:1573-6. [PMID: 23843132 DOI: 10.1007/s00247-013-2749-3] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 12/19/2012] [Revised: 05/12/2013] [Accepted: 05/23/2013] [Indexed: 11/30/2022]
Abstract
BACKGROUND Two types of Blake chest drains are used by our cardiac surgeons for management of their patients after thoracic surgery. The drain can be longitudinally split by the surgeon resulting in a limb of the drain in each thoracic cavity. A split flat Blake drain has two radiopaque limbs and a split round Blake drain has a radiopaque limb and a less radiopaque limb. OBJECTIVE To describe the radiographic appearance of these drains and promote their accurate radiologic identification and description. MATERIALS AND METHODS We conducted a retrospective review of the ability of our radiologists to correctly identify the radiographic appearance of the two devices. We identified 48 cases; 30 contained two radiopaque limbs and 18 had a radiopaque and a less radiopaque limb. RESULTS In 25 of the 30 (83%) cases the configuration of two radiopaque limbs was correctly identified; however, in 0 of 18 (0%) cases was the configuration of a radiopaque and a less radiopaque limb correctly identified. CONCLUSION By improving awareness of different Blake drain configurations and appearances this study aims to promote proper identification, accurate reporting, and reduced cognitive errors arising from Blake drain misidentification.
Collapse
Affiliation(s)
- Jean-Marc Gauguet
- Department of Radiology, Boston Children's Hospital, 300 Longwood Ave., Boston, MA, 02115, USA,
| | | | | | | |
Collapse
|