Rationale and Objectives
To evaluate the strengths and limitations of a rib-unfolding software in a polytrauma context.
Materials and Methods
Chest computed tomography (CT) examinations of 110 patients were reviewed for specific detection of rib fractures using: (1) transverse CT sections ± multiplanar reformattings (ie, the standard of reference), and (2) unfolded rib images reconstructed by the CT Bone Reading software with the possibility of rib analysis along their long axis and creation of standard orthogonal views in different orientations of any area suspected of fracture.
Results
The software provided complete reconstruction of the whole rib cage in 94 patients (85.5%) and partially incomplete reconstruction in 16 patients (14.5%). The percentage of ribs inadequately reconstructed was 1.5% (40 of 2640 ribs), mainly related to unfused epiphyses (13 of 40), costal hypoplasia (8 of 40), and vertebral fracture (6 of 40). The sensitivity and specificity in detecting rib fractures at a per-patient, per-rib, and per-costal arc level ranged from 0.73 to 0.84 and 0.99 to 1, respectively. At a costal arc level, the reader’s misinterpretations accounted for 67% (4 of 6) of false-positive and 24% (20/84) of false-negative results, and interpretive difficulties were encountered for single-cortex fractures or fractures at the extremities of the costal shaft.
Conclusions
An accurate diagnosis of rib fracture was achieved with the reading of unfolded rib images. In a polytrauma context, the evaluated system could facilitate rib analysis.
Introduction
Whereas computed tomography (CT) is a useful method for detecting pathological changes involving the ribs and adjacent structures, counting the ribs and thus, precisely localizing lesions, has always been considered as a time-consuming task on CT examinations, especially in the context of multiple fractures on adjacent ribs. Several methods have successively been proposed to overcome these difficulties, based on the recognition of anatomic landmarks. In the early 1990s, identification of the medial clavicle and the sternal angle on sequential CT examinations allowed easy recognition of the first rib and second costal cartilage, respectively, from which sequential counting of the other ribs could be undertaken. However, counting from the sternoclavicular joint is tedious for mid and lower rib lesions, and this method is not applicable for counting ribs on abdominal CT studies that do not have images of the entire rib cage. Another approach was then proposed by Kim et al. with a reference point at the xiphoid process . Owing to the presence of numerous anatomic variations in the attachments of costal cartilages to the proximal xiphoid, this method did not improve correct localization of rib lesions on abdominal CT . The introduction of volumetric CT scanning dramatically modified the detection of rib fractures as multiplanar reformatting (MPR) and volume rendering (VR) on whole-body acquisitions became accessible in the context of polytrauma . In the emergency setting, Alkhadi et al. demonstrated that VR had a high accuracy and was considerably faster than transverse imaging . However, these authors pointed out some limitations and possible pitfalls of VR that could hamper detection of subtle or undisplaced rib fractures.
In this context, great interest has recently been directed toward analysis of ribs on virtually rendered unfolded views of the ribs that provide automated recognition and numbering of right and left ribs. Applied to the reading of chest CT examinations in the context of blunt trauma and multiple myeloma before and after treatment , this software was found to improve the detection of rib fractures and osteolyses, respectively, with significantly reduced reading times compared to conventional standard multiplanar reformats. However, these investigations included the possibility of manual editing prior to the automated segmentation step or use of correction tools when the automatically assigned image set was considered diagnostically insufficient . In the specific context of polytrauma patients, Ringl et al. reported the need for user’s intervention in 38.6% of the examined patients . Although it was subsequently followed by successful rib segmentation in the majority of cases, the authors did not provide information on the causes of incomplete or incorrect rib segmentation after the initial postprocessing procedure. Moreover, they reported false-positive diagnoses of fractures in 6–12% of patients but they did not describe the reasons for these erroneous interpretations. Because precise knowledge of the strengths and limitations of a new tool is an important prerequisite prior to its clinical implementation, we undertook the present study to test the results achievable with the unfolding rib program applied without any user’s intervention. In a population of trauma patients, our goals were to assess the radiologist’s rate of detection of rib fractures and to describe the causes of software errors to optimize its utilization. No attempt was made to investigate its role as a complement or substitute to the traditional method of rib fracture assessment.
Materials and Methods
Study Population
Get Radiology Tree app to read full this article<
Chest CT Examinations
Get Radiology Tree app to read full this article<
CT Diagnosis of Rib Fractures
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Conditions of Image Interpretation
Get Radiology Tree app to read full this article<
Statistical Analysis
Get Radiology Tree app to read full this article<
Results
Population Characteristics
Get Radiology Tree app to read full this article<
Interobserver Agreement for the Reading of Unfolded Rib Images
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Software Performance
Segmentation Performance
Get Radiology Tree app to read full this article<
Diagnostic Performance
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
TABLE 1
Characteristics of Fractures Depicted by the Standard of Reference
Total Number of Fractured Arcs
n = 309 Fractures of the Posterior Arc
( n = 97) Fractures of the Lateral Arc
( n = 104) Fractures of the Anterior Arc
( n = 108) Unifocal, undisplaced fractures
( n = 217) 68 65 84 Unifocal, displaced fractures
( n = 74) 19 35 20 Bifocal, displaced fractures
( n = 16) 9 4 3 Bifocal, undisplaced fracture
( n = 2) 1 0 1
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
TABLE 2
Characteristics of Fractures Depicted by the CT Bone Reading Software
Total Number of Fractured Arcs
n = 231 Fractures of the Posterior Arc
( n = 62) Fractures of the Lateral Arc
( n = 87) Fractures of the Anterior Arc
( n = 82) Unifocal, undisplaced fractures
( n = 150) 39 49 62 Unifocal, displaced fractures
( n = 66) 15 33 18 Bifocal, displaced fractures
( n = 13) 7 4 2 Bifocal, undisplaced fracture
( n = 2) 1 1 0
CT, computed tomography.
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
TABLE 3
Diagnostic Performance of the Software for Rib Fracture Diagnosis
Sensitivity Specificity Detection of rib fracture Patient level 0.84 1 Rib level 0.77 0.998 Costal arc level 0.73 0.999Anterior arc__0.72__0.998__Lateral arc__0.84__1__Posterior arc__0.62__0.999 Detection of displaced rib fracture Patient level 0.92 1 Rib level 0.87 0.999 Costal arc level 0.86 0.999Anterior arc__0.83__0.999__Lateral arc__0.92__0.999__Posterior arc__0.79__1
Get Radiology Tree app to read full this article<
Causes of Software Errors
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Discussion
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
Get Radiology Tree app to read full this article<
References
1. Bhalla M., McCauley D.I., Golimbu C., et. al.: Counting ribs on chest CT. J Comput Assist Tomogr 1990; 14: pp. 590-594.
2. Kurihara Y., Nakajima T., Galvin J.: Counting ribs on chest CT scans: the easiest way. AJR Am J Roentgenol 1995; 165: pp. 487.
3. Kim S.J., Im J.G., Cho S.T., et. al.: Rib counting on CT using the sternal approach. J Comput Assist Tomogr 1993; 17: pp. 358-362.
4. Lee H.Y., Yoo S.M., Song I.S., et. al.: Counting ribs on CT by assessing costal attachments to the proximal xiphoid. J Thorac Imaging 2006; 21: pp. 284-287.
5. Linsenmaier U., Krötz M., Haüser H., et. al.: Whole-body computed tomography in polytrauma: techniques and management. Eur Radiol 2002; 12: pp. 1728-1740.
6. Begemann P.G., Kemper J., Gatzka C., et. al.: Value of multiplanar reformations (MPR) in multidetector CT (MDCT) of acute vertebral fractures. Do we still have to read transverse images?. J Comput Assist Tomogr 2004; 28: pp. 572-580.
7. Alkhadi H., Wildermuth S., Marincek B., et. al.: Accuracy and time efficiency for the detection of thoracic cage fractures. Volume rendering compared with transverse computed tomography images. J Comput Assist Tomogr 2004; 28: pp. 378-385.
8. Ringl H., Lazar M., Töpker M., et. al.: The ribs unfolded—a CT visualization algorithm for fast detection of rib fractures: effect on sensitivity and specificity in trauma patients. Eur Radiol 2015; 25: pp. 1865-1874.
9. Bier G., Mustafa D.F., Kloth C., et. al.: Improved follow-up and response monitoring of thoracic cage involvement in multiple myeloma using a novel CT postprocessing software: the lessons we learned. AJR Am J Roentgenol 2016; 206: pp. 57-83.
10. Homann G., Weisel K., Mustfa D., et. al.: Improvement of diagnostic confidence for detection of multiple myeloma involvement of the ribs by a new CT software generating rib unfolded images: comparison with 5- and 1-mm axial images. Eur Radiol 2015; 44: pp. 971-979.
11. Seuss H., Dankerl P., Cavallaro A., et. al.: Osteoblastic lesion screening with an advanced post-processing package enabling in-plane rib reading in CT-images. BMC Med Imaging 2016; 16: pp. 39.
12. Ludwig K., Schülke C., Diederich S., et. al.: Detection of subtle undisplaced rib fractures in a porcine model: radiation dose requirement—digital flat-panel versus screen-film and storage-phosphor systems. Radiology 2003; 227: pp. 163-168.
13. Cho S.H., Sung Y.M., Kim M.S.: Missed rib fractures on evaluation of initial chest CT for trauma patients: pattern analysis and diagnostic value of coronal multiplanar reconstruction images with multidetector row CT. Br J Radiol 2012; 85: pp. e845-e850.
14. Love J.C., Symes S.A.: Understanding rib fracture patterns: incomplete and buckle fractures. J Forensic Sci 2004; 29: pp. 1153-1158.
15. Battle C.E., Hutchings H., Evans P.A.: Risk factors that predict mortality in patients with blunt chest wall trauma: a systematic review and meta-analysis. Injury 2012; 43: pp. 8-17.
16. Sirmali M., Turut H., Topçu S., et. al.: A comprehensive analysis of traumatic rib fractures: morbidity, mortality and management. Eur J Cardiothorac Surg 2003; 24: pp. 133-138.
17. Alisha C., Ganjan G., Jyothi H.: Risk factors affecting the prognosis in patients with pulmonary contusion following chest trauma. J Clin Diagn Res 2015; 9: pp. OC17-OC19. Epub 2015, August 1
18. Omert L., Yeaney W.W., Protech J.: Efficacy of thoracic computerized tomography in blunt chest trauma. Am Surg 2001; 67: pp. 660-664.
19. Hammon M., Dankerl P., Tsymbal A., et. al.: Automatic detection of lytic and blastic thoracolumbar spine metastases on computed tomography. Eur Radiol 2013; 23: pp. 1862-1870.
20. Expert panel on Thoracic Imaging, Travis H., Kirsch J., et. al.: ACR (appropriateness criteria review): rib fractures. J Thorac Imaging 2014; 29: pp. 364-366.