Applied Radiology

Pulmonary embolism (PE) and deep venous thrombosis (DVT) are components of the same disease entity, accounting for 300,000 to 600,000 hospitalizations and 50,000 to 100,000 deaths a year in the United States. Since appropriate treatment is estimated to decrease the mortality by 30%, early and accurate diagnosis is essential. Until recently, many different imaging tests were used to evaluate pulmonary thromboembolic disease and thrombosis of the deep veins of the lower extremity. Ventilation perfusion scintigraphy (V/Q) has been the mainstay in the diagnosis of PE, with pulmonary angiography being the gold standard. However 40% to 70% of V/Q scans are nondiagnostic, thereby requiring additional testing for the diagnosis or exclusion of PE. Pulmonary angiography is an invasive test with a major complication rate of 1.3%. For the last 5 years, helical (spiral) CT has been used in the diagnostic evaluation of thromboembolic disease.

The advent of faster helical scanners, particularly multislice scanners, has revolutionized the diagnosis of not only clinically significant central pulmonary emboli, but has also greatly improved the detection of subsegmental pulmonary emboli. ^1-3 Faster scanning times, single breath-hold acquisitions, and thinner collimation has led to improved detection of both segmental and subsegmental pulmonary emboli with the diagnostic accuracy of CT pulmonary angiography (CTPA) approaching that of catheter angiography. ³ Therefore, CTPA is commonly used as a diagnostic modality of choice, especially in patients with an abnormal chest radiograph or underlying pulmonary disease, as both increase the frequency of a nondiagnostic V/Q scan. CTPA is also valuable as it yields an alternative diagnosis in 40% to 70% of patients with a negative PE scan, including bronchogenic carcinoma, metastatic disease, pericardial effusion, and aortic dissection. Recent studies demonstrate significantly improved visualization of segmental and subsegmental pulmonary arteries using thinner collimation. Remy-Jardin et al ¹ reported improved visualization of the segmental pulmonary arteries from 85% to 93% and that of the subsegmental vessels from 37% to 61% using the 2-mm technique compared with the 3-mm technique with single-detector CT scanner (SDCT). ¹ Patel et al ² recently reported an incremental improvement in visualization of the segmental arteries from 74% to 79% to 89%, respectively, when comparing the 3-mm collimation technique using SDCT with the 2.5-mm and 1.25-mm techniques using multidetector CT (MDCT). The incremental improvement was even better at the subsegmental levels, 35% with the 3-mm technique and 44% and 68% with the 2.5-mm and 1.25-mm techniques, respectively. ² Not only were arteries seen more clearly, but reader agreement as to presence or absence of thrombus in segmental and subsegmental vessels was more consistent with the 1.25-mm collimation technique. A recent large study compared the performance of dual-section helical CT with pulmonary angiography and concluded that the sensitivity, specificity, and positive- and negative-predictive values were 90%, 94%, 90%, and 94%, respectively. ³ They concluded that helical CT may replace pulmonary angiography in the diagnosis of PE and that selective pulmonary angiography should be reserved for patients with an unresolved diagnosis. ³

Ultrasonography (US), the most commonly used diagnostic tool for the evaluation of DVT, has a sensitivity of 95% and a specificity of 98% for the diagnosis of acute DVT. Currently, direct catheter venography, plethysmography, and isotope scanning are not often used. Recently, several investigators have compared US and helical CT in patients with suspected PE. Some authors advocate US to exclude a DVT in the event of a negative CTPA, thereby excluding clinically significant pulmonary embolism and the need for unnecessary anticoagulation, which has its own associated morbidity. However, this is at the expense of two different tests that take longer to perform, not to mention the additional costs. With the advent of newer and faster helical CT scanners, in particular multislice scanners, a single comprehensive exam can be performed to evaluate the pulmonary arterial bed and the deep veins of the pelvis and thighs. In a study of 71 patients, Loud et al ⁴ demonstrated that CT venography (CTV) is comparable with US in the evaluation of DVT with a sensitivity and specificity of 100%. Garg et al ⁵ reported obtaining a good quality CTV examination in 97% of a total of 70 patients. They demonstrated that CTV was more efficacious than US in 36% of cases, equivalent to US in 37%, and US was better than CTV in 27%. Duwe et al ⁶ found an accuracy of 93% for CTV using US as the reference test in 74 patients, with a sensitivity of 89%, specificity of 94%, positive-predictive value of 67%, and negative-predictive value of 98%. The CTV techniques used by these authors differ, with the scan interval varying from 5 mm to 20 mm, collimation 5 mm to 10 mm, and the injection delay from 3 minutes to 3.5 minutes. All of these studies were performed on a single-detector helical scanner. CTV is particularly useful in postoperative patients or in obese and immobile patients in whom sonography can be challenging to perform.

Multi-detector CT Scan Protocol: Part 1-Thorax

For the thorax protocol, 150 cc of iohexol 300 mg/mL (Omnipaque ^® [iohexol]; Amersham Health, Princeton, NJ) is injected via a 20-gauge angiocatheter in an antecubital vein at a rate of 4 cc/sec. A scan delay of 25 seconds is used. Imaging is performed from the dome of the diaphragm to the aortic arch during a single breath-hold, with the acquisition time varying between 15 to 19 seconds, depending on the size of the patient. Images are obtained at 1.25-mm collimation and reconstructed at 50% of the scan collimation (HS mode, pitch 6).

Part 2-Leg/pelvic veins

Three minutes following start of the contrast injection, acquisition of the pelvis and lower extremity veins is obtained from a caudad to cephalad direction from the level of the tibial plateaus to the iliac crests. Images are obtained at 7.5-mm collimation and reconstructed at 50% of the scan collimation (HS mode, pitch 6). The entire study is then evaluated on a GE Advantage Windows Workstation (GE Medical Systems, Milwaukee, WI). Use of the HS mode in large patients may account for very noisy images, limiting evaluation of the pulmonary arteries. In these patients, the HQ mode can be used and thicker slices obtained. Acquisition of the pulmonary arteries and the deep veins of the pelvis and thighs is obtained with a single bolus of intravenous contrast.

Study Interpretation

The window level and width is altered for optimum evaluation of the pulmonary arteries. A wider window width and a higher level may be needed to evaluate for subtle emboli. Depending on the concentration of the contrast in the pulmonary arteries, this may need to be altered at different levels. Then, the main, lobar, segmental, and subsegmental arteries are examined systematically from their origins to the 5th or 6th order branches, using the scrolling mode device. Pulmonary emboli are diagnosed as partial or complete filling defects, enlargement of the pulmonary artery with a large occlusive clot, peripheral filling defect, or nonenhancement of a segment of the pulmonary artery (Figure 1). Evaluation for chronic pulmonary emboli is also possible, with the presence of webs or detection of small thick-walled pulmonary arteries with enhancing walls and filling defects. Systematic review of the entire pulmonary arterial circulation is performed bilaterally. Note is also made of additional findings such as pleural or pericardial effusions, nodules, tumors, nodal enlargement, consolidation, emphysema, and aortic dissection or aneurysm. These aid clinicians in further management as the clinical presentation of the above diagnoses can be similar. Then, the veins of the pelvis and thighs are evaluated using the scroll mode device, after narrowing the window width and level, for optimum evaluation of the veins.

DVT is diagnosed when there is a central filling defect, enlargement of the vein (pitfall, venous valves), segmental nonvisualization, wall enhancement, or perivenous subcutaneous edema (Figures 2 and 3). Prolonged arterial enhancement may represent extensive bilateral DVT. However this can also be found when the CT is performed too soon in the arterial phase or when there is severe bilateral arterial insufficiency. Observation of other pathology is also possible, such as pelvic tumor compressing the pelvic veins or a Baker's cyst. Optimum venous enhancement is not uniform in all patients with an injection-to-scan delay of 3 minutes, as this varies according to the patient's cardiac status and the state of the pelvic and lower limb arteries. Patients with altered inflow due to atherosclerotic disease may require a longer delay time.

Multislice CT allows acquisition of 4 to 8 axial images with each revolution of the helical scanner, hence accounting for the short scan acquisition time, which is essential in sick, dyspneic intensive-care patients who cannot hold their breath. Not only has multislice CT improved evaluation of the segmental and subsegmental pulmonary arteries, but it has also allowed evaluation of the lower extremity veins, the source of PE, with a single injection of intravenous contrast. This translates into a single imaging modality for a quick, accurate, and efficient diagnosis of pulmonary thromboembolic disease. Recent studies have demonstrated <1% incidence of fatal PE in the event of a negative CTPA. ⁷ Currently, a multicenter trial, the Prospective Evaluation of Pulmonary Embolism Diagnosis II (PIOPED II) is underway to evaluate the accuracy of multislice CT in the diagnosis of PE and DVT.

REFERENCES

1. Remy-Jardin M, Remy J, Artuad D, et al. Peripheral pulmonary arteries: Optimization of the spiral CT acquisition protocol. Radiology. 1997;204:157-163.

2. Patel S, Kazerooni EA, Gross BH. Optimization of small pulmonary artery visualization for pulmonary embolism detection with multidetector CT [abstract]. Radiology. 1999;213(P):471.

3. Qanadli SD, Hajjam ME, Mesurolle B, et al. Pulmonary embolism detection: Prospective evaluation of dual-section helical CT versus selective pulmonary angiography in 157 patients. Radiology. 2000;217:447-455.

4. Loud PA, Katz DS, Klippenstein DL, et al. Combined CT venography and pulmonary angiography in suspected thromboembolic disease: Diagnostic accuracy for deep venous evaluation. AJR Am J Roentgenol. 2000;174:61-65.

5. Garg K, Kemp JL, Wojcik D, et al. Thromboembolic disease: Comparison of combined CT pulmonary angiography and venography with bilateral leg sonography in 70 patients. AJR Am J Roentgenol. 2000;175:997-1001.

6. Duwe KM, Shiau M, Budorick NE, et al. Evaluation of the lower extremity veins in patients with suspected pulmonary embolism: Retrospective comparison of helical CT venography and sonography. AJR Am J Roentgenol. 2000;175:1525-1531.

7. Svenson SJ, Sheedy PF, Ryu JH, et al. Clinical outcomes of patients with suspected acute pulmonary embolism after a negative CT scan. Presented at the 23rd Annual Scientific Session of the SCBT/MR, March 2000.

TECHNICAL ISSUES

Abdominal CT Angiography in the Era of Multidetector CT

Marc J. Fenstermacher, MD; Silvana Faria, MD; Paul M. Silverman, MD

As new techniques have emerged over the last 15 years, the imaging evaluation of the vessels of the abdomen has undergone considerable evolution. Prior to the advent of multidetector CT (MDCT), much of the advancement has been in magnetic resonance angiography (MRA). A number of innovative MRA techniques have been utilized, including 2-dimensional (2D) and 3-dimensional (3D) time-of-flight (TOF), phase contrast, and, most recently, bolus contrast-enhanced 3D fast gradient echo techniques, with or without step-wise incremental couch movement. Currently, the latter approach is widely utilized in most practices, especially for the evaluation of the arterial and portal venous systems. The systemic venous system remains a challenge for accurate, high-contrast MRA, primarily because of the relatively low concentration of injected gadolinium once it reaches the iliac veins and inferior vena cava. All of the MRA techniques in the abdomen share a number of drawbacks, which have fueled the search for better noninvasive angiographic techniques. With MRA, it is not possible to demonstrate both the vasculature and the regional soft tissues in the same imaging sequence; the highest quality arterial MRA will demonstrate bright vessels and practically nothing else. This problem is particularly limiting in the oncologic setting, where noninvasive angiography is used primarily to stage tumors locally and in surgical planning. Even high-quality abdominal MRA techniques still suffer relatively low spatial resolution due to the requirement of imaging in the coronal plane for adequate coverage, with a large field-of-view, yet avoiding wrap-around aliasing artifacts. Tertiary or smaller branches are not visualized due to limitations in resolution and slice thickness. Any patient motion during the 20- to 30-second breath-hold acquisition degrades the MRA images severely.

With single-detector helical CT (SDCT), it first became possible to reconstruct the 3D helix retrospectively with the acquired slice thickness but with much smaller intervals, such as 50% of the collimation. This enabled the creation of high-quality multiplanar volume reformations and other angiographic images using a workstation, and we began using this technique in selected cases for visualizing the arterial and venous vasculature adjacent to tumors in the abdomen and pelvis. The limiting factors with SDCT have been the interrelated issues of slice thickness and coverage. All bolus-contrast-enhanced angiographic techniques depend on matching the timing of image acquisition to the transit of contrast through the vessels of interest. Since all acquisitions are in the axial plane, there is limited coverage with the 2- to 3-mm collimation and pitch of 2 usually used with SDCT. ¹

The advent of MDCT has allowed the acquisition of larger areas of coverage with thinner slice collimation and much faster imaging times per series. We routinely image the entire abdominal aorta and proximal common iliac arteries in <20 seconds with 1.25-mm collimation. This is repeated during multiple phases of contrast enhancement when indicated, so that arterial, portal venous, and delayed venous phases may be acquired. When reconstructed with an interval of 0.63 mm, and transferred to a workstation for angiographic image creation, the result is very high-quality curved and multiple oblique reformations depicting the anatomy and pathology of interest optimally (Figure 1). Pertinent vessels are demonstrated, the adjacent soft tissues are well-depicted and their relationship to vessels visualized. In the oncologic setting, tailoring the examination to each clinical situation allows the radiologist to provide the angiographic images needed by the clinician as well as the appropriate phase(s) of contrast-enhanced axial images for evaluating other organs, such as the liver. Our current technique on a GE LightSpeed QX/I unit (GE Medical Systems, Milwaukee, WI) includes the following parameters: 1.25 mm thick, slice pitch 6 (beam pitch 1.5), 0.8 second rotation speed, interval 0.63 mm, kV 120, mAs 340. The slice collimation can be increased to 2.5 mm, if necessary, for coverage. SmartPrep (GE Medical Systems) is utilized, with a region of interest over the upper abdominal aorta, to time the arterial phase as close as possible to the transit of the bolus of contrast through the aorta. An alternative to SmartPrep, which does incur a nominal though mandatory delay between contrast arrival in the abdomen and the initiation of scanning, is to perform a test injection in order to estimate the contrast travel time to the abdominal aorta. This step is time consuming and not used routinely in our department at this time.

Data is transferred to a workstation, currently a Vital Images Vitrea2 (Vital Images, Inc., Minneapolis, MN) or an Advantage Windows workstation (GE Medical Systems), for creating and filming curved and multiple oblique reformations. Although this is a time-consuming step for the radiologist, it is rewarding for the clinician. While a technologist with special interest and training in image processing can perform this step in many cases, optimizing reformatted imaging planes often requires a knowledge of anatomy and pathology that may be beyond their reach (Figure 2). At the current stage of development of these techniques, interested cross-sectional radiologists should be encouraged to perform the reformations and reconstructions personally. The choice of optimum angiographic display (shaded surface display, MIP reconstruction, curved reformation and/or multiple oblique reformation) should be individualized for each case.

Recently published studies have begun to document the utility and accuracy of MDCT angiography. ^2,3 Although the techniques are still under development and systematic evaluation, this method of noninvasive angiographic imaging promises to supplant MRA for many applications.

REFERENCES

1. Kaatee R, Beek FJA, de Lange EE, et al. Renal artery stenosis: Detection and quantification with spiral CT angiography versus optimized digital subtraction angiography. Radiology. 1997;205:121-127.

2. Rubin GD, Shiau MC, Leung AN, et al. Aorta and iliac arteries: Single versus multiple detector-row helical CT angiography. Radiology. 2000; 215:670-676.

3. Matsumoto A, Kitamoto M, Imamura M, et al. Three-dimensional portography using multislice helical CT is clinically useful for management of gastric fundic varices. AJR Am J Roentgenol. 2001;176: 899-905.

PRACTICAL ISSUES

Imaging Considerations in Pediatric Multislice CT: A Radiologist's Perspective

Farzin Eftekhari, MD and Paul M. Silverman, MD

Multislice CT, occasionally referred to as multidetector or multidetector row CT, has made a significant impact on the imaging of pediatric patients by reducing the total study time in uncooperative or very ill patients. In many instances, this has virtually eliminated breathing or motion artifacts, and, even more importantly, the need for conscious sedation or general anesthesia. ¹

In those patients who require conscious sedation or general anesthesia, shorter examination time has minimized the sedation period, and, thus, the potential risks of sedation or anesthesia. This has increased patient throughput and the efficiency of the sedation team, enabling them to schedule additional pediatric CT, magnetic resonance (MR), or interventional radiology procedures.

Shortly after their release, several reports indicated that GE LightSpeed multislice systems (GE Medical Systems, Milwaukee, WI] delivered more radiation to patients when properly compared (apples-to-apples) with single-slice CT scanners. ^2-4 With subsequent deployment of multislice scanners, a sense of unease has arisen in utilizing these scanners for pediatric imaging. This presents a significant quandary since their increased speed makes them highly desirable for pediatric imaging, while any increased radiation dose prompts concerns regarding the long-term effects of radiation in young patients. ^5-7 In August 2000, we began routine imaging of pediatric patients with our multislice units (GE LightSpeed). We initially decreased the mA and increased the table speed (relative to the CT/i platform [GE Medical Systems]) to minimize overall radiation dose and improve image quality by reducing motion artifacts. In February 2001, anthropomorphic pediatric phantoms were acquired and will be used to refine pediatric CT protocols with respect to radiation dose.

Our current technique is shown in Table 1, and we hope for further decrease in radiation dose in the future.

Our initial calculations for multislice scanners (Table 2) indicated a two- to four-fold increase in the dose compared with conventional helical scanners when tested at identical kV and mA-seconds per scan. Note that dose increased as collimation was reduced. In the first two lines of Table 2, the 5-mm single-slice conventional result was 0.8R; and 3.1R for the multislice unit prior to an early upgrade, then fell to 1.9R after the focal spot tracking upgrade was installed. The LightSpeed multislice scanner upgrade installed in the summer of 2000 allowed compensation for focal spot motion to be better controlled. This approach better limits the radiation that is not effectively used by the multidetector system. For 10-mm collimation, the single-slice result was also 0.8R; and 2.1R for the multislice unit prior to the upgrade. This dose then dropped to 1.3R after the upgrade. This relatively increased dose on the LightSpeed 5-mm compared with the 10-mm slice thickness was because the collimator must be set to produce a uniform fan beam of radiation across all of the detectors in the z-axis direction. This, in turn, requires some "spill-over" of radiation beyond the active detector surface, which increases the dose proportionally higher in thinner-slice acquisitions. This is especially important to note in young patients, who often have thinner slices used to better define pathological processes and to cover the smaller anatomy.

It is also important to realize that the source-to-detector distance is decreased in the LightSpeed multislice scanner (94.9 cm) relative to the GE single-slice design (109.9 cm). This detector alone accounts for a 34% increase in radiation dose for the multislice system compared with the single-slice configuration using identical scan parameters.

For pediatric abdominal/pelvic CT scanning, we use a slice thickness of 5 mm, high speed (HS), and a table speed of 22.5 mm/rotation. For pediatric chest CT, we use a slice thickness of 5 mm, HS, and 22.5 mm/rotation table speed.

Pediatric patients receive iodinated contrast material when it is clinically indicated. To limit reactions or nausea and vomiting, we use low-osmolality contrast material (LOCM) at a dose of 1.5mL/kg body weight for the chest, and 2.0 mL/kg for abdominal and pelvic CT. We prefer to deliver all of the contrast before scanning and allow a 25-second delay for chest, and chest/abdomen/pelvis imaging (allowing a 12-second pause between chest and abdomen), and 50 seconds for abdomen/pelvis imaging. Alternatively, when available, we use a computer automated scanning technology, such as SmartPrep, to ensure that scanning occurs during peak liver enhancement. This allows for consistency and optimal imaging to detect focal liver lesions.

Injection rates for contrast material range from 0.3 to 3.0mL/sec, depending on the size of the IV access line (22- to 18-gauge angiocatheters, depending on the age of the child).

Our pediatric technique (Table 1) is based on the child's body size (depth of field of view), age, and weight. For children older than 6 years of age, we use a slice thickness of 7.5 mm and 15 mm/rotation table speed. We use this table to select the appropriate mA setting for each case. The mA ranges from 140 to 230 with a kV of 120. This table is designed for current GE multislice scanners in which the scan time is 0.8 seconds. However, since newer scanners offer shorter scan times, such as 0.5 seconds, we intend to update this table to adjust for mA values accordingly.

Conclusion

We feel that the use of multislice CT scanners has had a significant impact on imaging in pediatric patient population in terms of need for sedation and throughput, but we need to continue to look for ways to decrease the radiation dose to even lower levels.

Acknowledgments

The authors would like to thank Dianna D. Cody, PhD, and Donna M. Moxley, MS, from Diagnostic Physics for their valuable comments.

REFERENCES

1. Pappas JN, Donnelly LF, Frush DP. Reduced frequency of sedation of young children with multisection helical CT. Radiology. 2000;215:897-899.

2. McCollough CH, Zink FE. Performance evaluation of a multislice CT system. Med Phys. 1999;26:2223-2230.

3. Cody DD, Moxley DM, Eftekhari F, Silverman PM. Pediatric dose considerations in multislice helical CT. Helical CT Today. 2001;6(3):3-4.

4. Motley DM, Hazle JD, Shepard JS, Zhou XJ. Dosimetry of a new multislice helical CT scanner: Calculated and measured patient doses [abstract]. Radiology. 1999;213(P):284.

5. Donnelly LF, Frush DP, Nelson RC. Multislice helical CT to facilitate combined CT of the neck, chest, abdomen, and pelvis in children. AJR Am J Roentgenol. 2000;174:1620-1622.

6. Brenner DJ, Elliston CD, Hall EJ, Berdon WE. Estimated risks of radiation-induced fatal cancer from pediatric CT. AJR Am J Roentgenol. 2001;176:289-296.

7. Paterson A, Frush DP, Donnelly LF. Helical CT of the body: Are settings adjusted for pediatric patients? AJR Am J Roentgenol. 2001;176:297-301.

MEDICAL ECONOMICS

Application of Six Sigma to Healthcare: Improving CT Efficiency

Steve Venable, MBA and Paul M. Silverman, MD

Six Sigma is a performance improvement methodology used to achieve significant change and improvement. The term Six Sigma refers to six standard deviations from the mean on a normal distribution model. This level of performance would seem impossible to attain at first, but the methodology leads to extraordinary improvement on processes and final results. Many large corporations, including Allied Signal, Honda, and GE, ¹ have adopted Six Sigma to improve their financial performance and maintain a competitive advantage.

Six Sigma represents a cultural shift to continuous, data-driven performance improvement. Though Six Sigma was originally used to decrease defects in manufacturing industries, it has moved to service industries, and is readily adapted to healthcare. The data-driven aspect of Six Sigma meshes well in the current hospital setting where dramatic efforts are being made to strive for marginal efficiency in a competitive environment. Physicians require a business methodology based on verifiable data and analysis. This focuses all groups on solutions to problems and quantitative measures of success.

At the University of Texas M.D. Anderson Cancer Center (MDACC), we have used Six Sigma to provide a dramatic improvement in efficiency. In computed tomography (CT), a common reason given for patient delays was a lack of laboratory tests prior to the CT scan. More than 95% of our patients received IV contrast and labs were required prior to administering contrast. However, data analysis found that <3% of patients needing IV contrast arrived in CT without labs available (Figure 1). Today, the staff address issues by collecting data first before stating the reasons for failure.

Training is a key component in Six Sigma. All levels in the organization must undergo training in the methodology and tools used. At MDACC, the executive staff underwent 3 days of training at a high level, the division leadership had 5 days, and the team participants had 10 days. The commitment in resources and time is substantial, but the payoff is extraordinary. Teaming with an outside consulting group versed in Six Sigma methodology is critical. After several projects are completed, the reliance on outside services can be reduced.

The heart of the Six Sigma Methodology is the DMAIC process. ^1,2 There are several acronyms but this reflects the gist of most: Define the key objectives; Measure the baseline, benchmark the overall process; Analyze the various processes performance; Improve the process, measure for desired affect; Control the improved process.

The first task is to define or characterize the process and problems. This provides the benchmark against which the improvements can be measured. We discovered that many long-held perceptions were inaccurate. Perception : "Many" patients reported without the requisite labs; Reality : <3% were missing laboratory tests. Perception : CT scanners were a bottleneck to efficient patient scanning; Reality : CT scan time averaged 3 minutes per patient, the shortest cycle time measured.

Why in CT?

At MDACC we suffered from limited CT capacity (with patients going elsewhere for CT studies), high CT technologist turnover, long patient wait times, and a stressful work environment. Several performance initiatives had been completed through Diagnostic Radiology prior to Six Sigma, but none had addressed the fundamental problems. The challenge was to address these issues successfully.

The patient process was analyzed and broken into several subprocesses to understand the cause of patient delays. The result was a 75% reduction of the average patient wait for a CT exam, from 1 hour to 15 minutes. It was realized that two distinct processes were taking place, one relating to the patient and one of interpreting the studies. Subprocesses such as scheduling, patient prep, and the CT exam were driven by the patient, not by the exam or its type. The subprocesses of interpretation and report generation were driven by the number of individual procedures performed, not by the patient.

CT scan time did not limit capacity, the processes leading to and following the CT scan were the bottlenecks. The CT scanners were idle for many minutes per hour. The key to improving CT capacity was to identify and remove the bottlenecks. This reduced patient wait times and improved patient satisfaction, reducing staff stress by reducing patient complaints. MDACC moved from performing 207 procedures per day on 6 CT scanners to 272 procedures per day, a 31% increase on the same installed scanner base. Today, we average 320 procedures per day on 6.5 scanners.

The actual time the CT scanner is dedicated to a patient was 23 minutes, regardless of the type or number of CT exams performed on a patient. This was due to use of high-speed helical scanners, with scan time never >3 minutes, even for a combined chest, abdomen, and pelvis exam. We spent more time getting the patient onto and off the table, and the setup and cleaning of the room, 20 minutes total. The team concentrated on optimizing the patient-related activities and removed the excess idle time of the CT scanners.

The Interpretive Process

The teams also analyzed the activities in the reading rooms. The goal was to eliminate all nonvalue activities performed by the radiologists and shift those activities to others, allowing the radiologists to concentrate on film interpretation and reports. We began to print out the information needed by the radiologists for protocoling the patients putting pertinent information at their fingertips. We also began to hang the motorized viewers continuously throughout the day, relieving the radiologist of this tedious task. The results were a dramatic increase in throughput in the reading rooms and the radiologists were able to stay current with the growing patient activity.

During the characterization process, several processes were determined