Comparison of MIP and VR With Average at Various Slab Thickness in MDCT For Detecting Pulmonary Nodules.

Sharma D

ISSN 0972-5997

Published Quarterly

Mangalore, India

editor.ojhas@gmail.com

Custom Search

OJHAS Vol. 15, Issue 3: (July-September 2016)

Original Article
Comparison of MIP and VR With Average at Various Slab Thickness in MDCT For Detecting Pulmonary Nodules

Authors
Dolly Sharma, Assistant Professor, Department of Medical Imaging Technology, School of Allied Health Sciences, Manipal University, Manipal,
Sushil Yadav, Senior Scale Assistant Professor, Department of Medical Imaging Technology, School of Allied Health Sciences, Manipal University, Manipal,
Prakshini K, Associate Professor, Department of Radio Diagnosis and Imaging, Kasturba Medical College, Manipal,
Sameer Paruthikunnan, Senior Resident, Department of Radio Diagnosis and Imaging, Kasturba Medical College, Manipal,
Winniecia Dhkar, Assistant Professor, Department of Medical Imaging Technology, School of Allied Health Sciences, Manipal University, Manipal, India.

Address for Correspondence
Dolly Sharma,
Assistant Professor,
Department of Medical Imaging Technology,
School of Allied Health Sciences,
Manipal University,
Manipal, India.
E-mail: sdolly468@gmail.com

Citation
Sharma D, Yadav S, Prakshini K, Paruthikunnan S, Dhkar W. Comparison of MIP and VR With Average at Various Slab Thickness in MDCT For Detecting Pulmonary Nodules. Online J Health Allied Scs. 2016;15(3):6. Available at URL: http://www.ojhas.org/issue59/2016-3-6.html

Open Access Archives
http://cogprints.org/view/subjects/OJHAS.html
http://openmed.nic.in/view/subjects/ojhas.html

Submitted: May 12, 2016; Accepted: Oct 3, 2016; Published: Oct 25, 2016

This work is licensed under a Creative Commons Attribution-No Derivative Works 2.5 India License

This Article

Abstract: Introduction: Lung cancer found to be the most common cause of death worldwide. Early detection of pulmonary nodules can improve the survival rate of the patients suffering from lung cancer. In case of patients with known pulmonary metastasis, infection, other diseases the presence of bilateral pulmonary nodules is an indicative of metastasis. Objective: to compare the detection rate of pulmonary nodules in MIP and VR and to find out best slab thickness. Materials and methods: 15 patients referred for HRCT who had <20 nodules in each lung on source (1mm) images and who did not have any evidence of fibrosis in lungs were included. The source images were reconstructed into MIP and VR (slab thickness: 4, 7 and 11 mm; Interval: 3.5mm). Two radiologists independently evaluated different thickness of MIP and VR in separate sittings and documented the number of nodules with size and density of each nodule. Difference between the number of nodules detected on MIP and VR at various thickness was assessed using Wilcoxon sign rank test and p values were calculated to assess the significance. Results: Both readers detected more nodules on VR as compared to MIP (Total nodules on VR: 493, Total nodules on MIP: 467) with p- value being 0.262, indicating that there was no statistical significant difference between both techniques. Slab thickness of 11mm was best for nodule detection on MIP (p value: 0.027); however, there was no statistically significant difference between 7 mm and 11 mm thickness on VR (p: 0.3). 11 mm MIP was also better than the other thickness in detection of low density nodules (p: 0.039); but there was no significant difference in the detection of high density nodules (p: 0.131). Conclusion: MIP and VR were equally accurate in detecting lung nodules. 11mm MIP detected significantly more number of nodules than 7 mm, but both thickness were equally accurate on VR.
Key Words: Computed tomography, Multi slice computed tomography, Maximum intensity projection, Volume rendering

Introduction:

Early detection of pulmonary nodules can improve the survival rate of the patients suffering from lung cancer. The distribution, size and density of small nodules is generally of more significant in differential diagnosis, so all the features are usually taken into account for diagnosis.¹ CT is most sensitive in detection of pulmonary nodes,²usually recommended to follow the changes in nodule size, number and morphology.^3,4 CT has post processing tools to reduce the fatigue and time of the observer for nodule detection these tools include Maximum Intensity Projection (MIP) and Volume Rendering (VR).⁵ The basic clinical application of MIP is to improve the detection of pulmonary nodules and the assessment of their profusion. MIP makes also easier the differentiation between ground glass opacity and mosaic perfusion in case of mosaic attenuation pattern. Volume Rendering technique, which provides a color and opacity to individual density, may result in an image close to the macroscopic pathological view, with promising perspectives.⁶ VR is less user dependent, measurements particularly in those structures perpendicular to the axial plane are more accurately obtained.⁷ We compared MIP, VR with Average at various slab thicknesses for detecting pulmonary nodules in MDCT. Objectives of the study was to compare MIP with VR technique in detecting pulmonary nodules. To ascertain the best slab thickness of MIP and VR (4, 7 and 11mm) and to know the effective reconstruction method for different nodule size and density. As per our knowledge any study had not been done in past, about the detection rate with respect to the density of nodules so, this study would reflect the importance of reconstruction technique in low and high density nodules.

Materials and Methods:

A cross sectional prospective study was conducted by non-probability sampling technique over a period of 6 months after obtaining institutional ethical clearance. All the patients referred to radiology department with history of metastatic or infective lung diseases and who underwent HRCT thorax were included. The age group ranged from 18-81 years. For the diversities of nodules the data set with nodules <45 in both lungs was only used. Patients with fibrosis, scaring, pneumonia/atelectasis and motion artefact were excluded from the study. Patients on ventilator support were excluded from the study.

HRCT scan of thorax was performed on MDCT Brilliance 64 slice Phillips with a scan time of 4.341 sec at an interval of 10mm, from the apex of the lungs to the diaphragm using collimation of 64 x 0.625 and slice thickness of 1mm with a lung enhanced filter for producing standard resolution.

During the scan, the patient was instructed to breathe at full inspiration and hold his /her breath for a few sec during scanning time in-order to reduce motion artefacts.

The original series of HRCT scan having thickness of 1mm with incrimantation of 10mm were reconstructed into thin sections with a thickness of 1mm and incrementation of 0.5mm using a high resolution reconstruction algorithm with standardized window level and width settings for lung parenchyma (window level, -650 HU; window width,1500 HU). Scanning length varies between 30 and 35cm and resulted in 300-350 transverse images per patient data set included in the study sent to a dedicated workstation Philips Extended Brilliance workspace with Philips CT viewer software allows real time 3D processing of MIP and VR reconstructions without user interaction.

Average data set with slab thickness of 1mm with 0.5 interval was included in the study. Nodules characteristics density and size were defined. MIP and VR images at 4, 7, and 11mm slab thicknesses with 3.5mm incrementation were reconstructed from 1mm transverse sections; MIP images were displayed at a Window center of -300 HU and window level of 1600 HU. Nodules density was defined only on MIP images data sets with a threshold of -100 HU, nodules with less than -100 HU were low density and those more than 100 HU were high density nodules. VR images were observed at 49% opacity in bone 4 window.

Two radiologists independently analyzed the Average, MIP and VR data sets of different thicknesses as Reader 1 and Reader 2 who were experienced for 10 and 3 years, who were blinded for the number of lesions per patient. Both readers did not see MIP and VR images in a particular time within the same reading session to avoid memory effects, we ensured that at least 2 weeks elapsed between readings. Maximum diameter of each nodule detected by both readers on MIP and VR data sets was measured and classified according to their sizes into <3mm, 4-6mm and 6-10mm. Doubtful lesions were confirmed by consensus after seeing the multiple planes. Data collected for this study was analyzed using median and inter quartile range to summarize the minimum and maximum difference in nodule count between MIP and VR. Wilcoxon sign rank test is used to test the median difference in nodule count. A P value near to 1 suggests that the median difference in nodule count is zero between MIP and VR. All statistical analysis was carried out using SPSS 16.

Fig 1 (a,b): Coronal planes of MIP and VR showing detected nodules

Fig 2 (a,b,c): MIP images reconstructed at a. 4mm, b. 7mm and c. 11mm slab thickness showing the increase in number of pulmonary nodules with increase in slab thickness.

Fig 3 (a, b, c): VR images reconstructed at a. 4mm, b. 7mm and c. 11mm slab thickness showing the increase in number of pulmonary nodules with increase in slab thickness.

Fig 4 (a, b): comparison of VR and MIP with respect to difference in the number of nodules detected at same slab thickness a. VR image at 1mm slab thickness, b. MIP image at 11mm slab thickness

Results:

Total 65 patients underwent HRCT for different indications were included. 32 out of these 65 were found to have fibrosis and 12 patients had nodules >45 in both lungs and 6 had motion blur hence excluded. Thus 15 patients who met the inclusion criteria were finally included for the nodule assessment.

Table 1: Comparison of nodule count with respect to MIP and VR by readers

Total nodule count
Minimum Difference

Maximum Difference

P value

MIP

VR

MIP

VR

MIP

VR

464

492

1

1

464

492

0.262

Table 1 shows the total number of nodules detected by MIP and VR technique by two radiologist. It was found that P value was not statistically significant for number of nodules detected by MIP and VR as both could pick up almost similar number. The inter-reader variability was not perceived as there was very good statistical agreement between both readers (kappa value).

Table 2: Comparison of nodule count with respect to MIP and VR at various slab Thickness by readers
Slab thickness	Total nodule count		Minimum difference		Maximum difference
	MIP	VR	MIP	VR	MIP	VR
4mm	281	451	1	1	40	94
7mm	390	444	1	1	82	76
11mm	463	474	1	1	85	97

Total nodule count varied significantly at different slab thickness that is 4, 7 and 11 mm of MIP and VR.

Table 3: Comparison of nodule count with respect to MIP and VR at various slab thickness by readers

Slab thickness

Total nodule count

Minimum difference

Maximum difference

P value

MIP

VR

MIP

VR

MIP

VR

4mm

281

451

1

1

40

94

0.013

7mm

390

444

1

1

82

76

0.135

11mm

463

474

1

1

85

97

0.646

Table 3 shows that there was statistical significance between MIP and VR of 4mm slab thickness in nodule detection with P value of 0.013. However, there was no statistical difference between the two techniques with 7 and 11mm slab thickness with P value of > 0.05 and between radiologists also there was good agreement.

Table 4: Comparison of nodule count with respect to MIP 7 and 11mm MIP slab thickness by readers

MIP 7 and 11mm slab thickness

Total nodule count

Maximum

Minimum

Standard deviation

P Value

7mm

390

82

1

24.547

0.027

11mm

463

85

1

30.787

Table 4 gives the nodule count difference at 7 and 11mm MIP slab thickness where 11mm slab thickness was best with maximum and minimum difference 85 and 1 with standard deviation of 30.787 and a P value of 0.027.

Table 5: Comparison of nodule count with respect to VR at 7 and 11mm VR slab thickness by readers
VR 7 and 11mm slab Thickness	Total nodule count	Maximum	Minimum	Standard deviation	P value
7mm	451	76	1	26.293	0.101
11mm	474	97	1	30.561

Table 5 gives the nodule count difference between 7 and 11mm VR slab thickness however, there was no statistically significant difference between 7 mm and 11 mm thickness on VR with a P value of 0.101.

Table 6: Comparison of nodule count with respect to size at 11mm MIP and VR slab thickness by readers

Slab thickness

Nodule sizes

Total nodule count

Minimum difference

Maximum difference

P value

MIP

VR

MIP

VR

MIP

VR

<3mm

373

231

0

0

82

57

0.041

11mm

4-6mm

75

161

0

0

26

48

0.047

7-10mm

22

68

0

0

11

29

1.144

Table 6 gives the summary of Total nodule count between MIP and VR with respect to various size of nodule studied. There was statistically significant difference between MIP and VR for nodules less than 3mm and 4-6mm size with a P value of 0.041 and 0.047 respectively. However there was no significant difference between MIP and VR for 7-10mm size of nodules (P-1.144).

Table 7: Comparison of nodule count with respect to densities at various slab thickness on MIP by readers
Slab thickness	Densities	Total nodule Count	Minimum difference in nodule count	Maximum difference in nodule count
4mm	LD	126	0	36
	HD	155	0	36
7mm	LD	390	0	46
	HD	249	0	82
11mm	LD	180	0	74
	HD	281	0	84

Table 7 gives the minimum and maximum difference in nodule count according to densities at three different slab thickness on MIP data. Maximum difference in LD and HD nodule count was observed in 11 mm MIP slab thickness with 74 LD nodules and 84 HD nodules.

Table 8: Comparison of nodule count with respect to high density at 7 and 11mm MIP slab thickness

7 and 11mm slab thickness

Total nodule count

Mean

Maximum

Minimum

Standard Deviation

P value

7mm

249

16.60

82

0

22.646

0.131

11mm

281

18.73

84

0

25.952

Table 9: Comparison of nodule count with respect to low density at 7 and 11mm MIP slab thickness

7 and 11mm slab thickness
Total nodule count

Mean

Maximum

Minimum

Standard Deviation

P value

7mm

141

9.40

46

0

13.097

0.039

11mm

183

12.20

74

0

19.847

Table 8 and 9 gives 11 mm MIP better than the other thickness in detection of low density nodules with a P value of 0.039, but there was no significant difference between slab thicknesses with respect to high density nodules P value 0.131.

Discussion:

Detection of small and low density lung nodules are challenging. MIP was considered to be the best reconstruction technique. Recent past computer aided detection also showed high sensitivity (CAD).⁸ Lung nodules of any size is important in a known case of malignancy with metastasis. In a high risk screening group it is important to give significance to nodule size as small as 4 mm or more, as risk of malignancy is observed with > 4 mm nodule. Such a small nodule needs follow up, to know the interval change. It is also well known that ground glass density nodules are also having higher risk of turning in to malignancy. There were many studies done to know the best technique to identify these small and low density nodules like CAD software,^9,10 and MIP reconstruction with different thickness and showed high sensitivity.¹¹ However there were only few studies till now which compared VR images with MIP of different slab thickness and interval in nodule detection sensitivity. Our study aimed to find out the best reconstruction technique and then the best slab thickness in nodules of various size and density. We found that nodule count was higher on VR than MIP for both observers, however the median difference carried out using Wilcoxon sign rank test for nodule count between MIP and VR was not statistically significant and p value was more than 0.05. This suggests that both techniques were equally efficient in detecting pulmonary nodules. Average reading time with MIP (5minutes) was significantly shorter than VR (10 minutes). This might be due to the fact that the readers were more familiar to MIP images.

In the study performed by Philips et al in 2007 in which VR was found to be superior when compared to MIP in detection of nodules. Number of nodules detected by using VR was greater in comparison to MIP as VR displays the entire volume of data and not only the maximum intensity voxels. Therefore the information provided by VR was more relevant than MIP. However VR was superior to MIP only for nodules smaller than 10mm, for nodules larger than 10mm significant difference was not found due to the fact that evaluation of lesions was less prone to visual perception errors.¹²

Another study conducted by Park EA et al using 1mm thin sections and 5mm MIP for nodule detection, concluded that MIP as a post processing technique improves the diagnosis of pulmonary nodules.¹³ Various studies performed earlier shown that the detection rates for pulmonary nodules using post processing techniques such as MIP and VR are superior to those achieved through conventional transverse sections.¹⁴

Our study showed that slab thickness of 11mm was best for nodule detection on MIP. However, there was no statistically significant difference between 7 and 11mm with VR. MIP and VR with slab thickness of 11mm performed better than other slab thickness. Increase in slab thickness reduces the number of images and makes it more prone to partial volume arteifacts.¹⁴ This was not seen in our study, since we had used a smaller reconstruction interval of 3.5 mm.

Other studies also showed that increase in slab thickness improves the detection rate of pulmonary nodules, Diederich et al compared 15mm and 30mm MIP and found 15mm MIP to be slightly superior.¹⁵ A study performed by Kawel et al concluded that MIP with a thickness of the sliding thin slab of 8 mm had a significantly higher sensitivity for nodule detection in chest CT than all other tested techniques: 5-mm MIP, 11-mm MIP, 5-mm VR, 8-mm VR, and 11-mm VR.⁵

In current study, MIP was found to be significantly better than VR for the detection of nodules less than 7 mm, but there was no significant difference between MIP and VR for the detection of nodules more than 7mm of size. Same result was concluded by Philips et al while comparing MIP and VR, readers found VR to be significantly superior to MIP in the detection of nodules less than 10mm.¹²

Eleven mm MIP was found to be better than the other thickness in detection of low density nodules, but there was no significant difference in the detection of high density nodules. Furthermore, a count of low and high density nodules was observed by both readers in 4, 7 and 11mm MIP slab thickness.

Nodule detection rate remain unaffected of their size on 8mm MIP.⁵Other studies comparing 5mm MIP⁸ and 10mm MIP⁹to conventional axial images found MIP to be better only for the detection of nodules less than 5mm. Valencia et al compared axial and coronal MIP images with standard axial 1mm and 5mm slices in the detection of pulmonary nodules and found that nodules less than 5mm in size were accurately detected by the axial MIP reconstructions and nodules larger than 5mm were equally detected on all the modalities, thus concluding that non-overlapping axial MIP reconstructions improve the accuracy in the detection of small pulmonary nodules.¹⁶

Limitations

In current study most of the patients were excluded due to clinical reasons and included patients were less to reach a strong conclusion, however the number of nodules detected were enough to carried out the study so, in future more studies should be performed by taking large sample of patients and nodules to get strong correlation.

Our study had no absolute reference standard so, the maximum number of nodules detected by both readers were considered as the standard of reference.

Conclusion

MIP and VR were equally accurate in detection of lung nodules. 11mm MIP detected significantly more number of nodules than 7 mm, but both thickness were equally accurate on VR. Increase in slab thickness led to increase in the detection rate of nodules especially on MIP. MIP slabs were good for the detection of nodules less than 6mm of size and VR was better for the detection of nodules more than 7mm of size. MIP thicker slabs were best in the detection of both low density and high density nodules.

References

Jankowski A, Martinelli T, Timsit JF, Brambilla C, Thony F, Coulomb M, Ferretti G. Pulmonary nodule detection on MDCT images: evaluation of diagnostic performance using thin axial images, maximum intensity projections, and computer-assisted detection. Eur Radiol. 2007;17:3148-3156.
Marten K, Grillhosl A, Seyfarth T, Obenauer S, Rummeny EJ, Engelke C. Computer-assisted detection of pulmonary nodules: Evaluation of diagnostic performance using an expert knowledge-based detection system with variable reconstruction slice thickness settings. Eur Radiol 2005;15:203-212.
MacMahon H, Austin JHM, Gamsu G, et al. Guidelines for management of small pulmonary nodules detected on CT scans: a statement from the Fleischner Society. Radiology 2005;237:395-400.
Brown MS, Goldin JG, Rogers S, et al. Computer-aided lung nodule detection in CT: results of large-scale observer test. Acad Radiol 2005;12:681-6.
Kawel N, Seifert B, Luetolf M, et al. Effect of slab thickness on the CT detection of Pulmonary nodules: use of sliding thin-slab maximum intensity projection and Volume Rendering. AJR 2009;192:1324-1329.
Beigelman-Aubry C. Post-processing and display in multislice CT of the chest. JBR-BTR, 2007;90:85-88.
Roy G. Essentials of Computed tomography2006, 1st edition, chapter 16; 148
Zhan P, Xie H, Xu C et al. Management strategy of solitary pulmonary nodules. J Thorac Dis 2013;5(6):824-829.
Wiemker R, Rogalla P, Zwartkruis A, Blaffert T. Computer-aided lung nodule detection on high-resolution CT data. Proc SPIE. 2002;4684:677-688. doi: 10.1117/12.467210.
Park EA, Goo JM, Lee JW, et al. Efficacy of computer-aided detection system and thin-slab maximum intensity projection technique in the detection of pulmonary nodules in patients with resected metastases. Invest Radiol. 2009;44:105-113.
Coakley FV, Cohen MD, Johnson MS, Gonin R, Hanna MP. Maximum intensity projection images in the detection of simulated pulmonary nodules by spiral CT. Br J Radiol 1998;71:135-140.
Peloschek P, Sailer J, Weber M, Herold CJ, Prokop M, Schaefer-Prokop C. Pulmonary nodules: sensitivity of maximum intensity projection versus that of volume rendering of 3D multidetector CT data. Radiology. 2007;243:561-9.
Park EA, Goo JM, Lee JW, et al. Efficacy of computer-aided detection system and thin-slab maximum intensity projection technique in the detection of pulmonary nodules in patients with resected metastases. Invest Radiol. 2009; 44:105-113.
Eibel R, Turk TR, Kulinna C, Herrmann K, Reiser MF. Multidetector-row CT of the lungs: multiplanar reconstructions and maximum intensity projections for the detection of pulmonary nodules [in German]. Rofo 2001;173:815-821
Diederich S, Lentschig MG, Overbeck TR, Wormanns D, Heindel W. Detection of pulmonary nodules at spiral CT: comparison of maximum intensity projection sliding slabs and single-image reporting. Eur Radiol 2001;11:1345-1350.
Valencia R, Denecke T, Lehmkuhl L, Fischbach F, Felix R, Knollmann F. Value of axial and coronal maximum intensity projection (MIP) images in the detection of pulmonary nodules by Multislice spiral CT: comparison with axial 1-mm and 5-mm slices. Eur Radiol 2006; 16:325–332.