Computer-aided diagnosis system versus conventional reading system in low-dose (&lt; 2 mSv) computed tomography: comparative study for patients at risk of lung cancer

Wang, Dong; Cao, Lina; Li, Boya

doi:10.1590/1516-3180.2022.0130.R1.29042022

Acessibilidade / Reportar erro

Brasil

Sao Paulo Medical Journal

Español English

Brasil

Español English

sumário « anterior atual seguinte »

Sumário

Original Article • Sao Paulo Med. J. 141 (2) • 2023 • https://doi.org/10.1590/1516-3180.2022.0130.R1.29042022 copy

Computer-aided diagnosis system versus conventional reading system in low-dose (< 2 mSv) computed tomography: comparative study for patients at risk of lung cancer

Authorship SCIMAGO INSTITUTIONS RANKINGS

ABSTRACT

BACKGROUND:

Computer-aided diagnosis in low-dose (≤ 3 mSv) computed tomography (CT) is a potential screening tool for lung nodules, with quality interpretation and less inter-observer variability among readers. Therefore, we aimed to determine the screening potential of CT using a radiation dose that does not exceed 2 mSv.

OBJECTIVE:

We aimed to compare the diagnostic parameters of low-dose (< 2 mSv) CT interpretation results using a computer-aided diagnosis system for lung cancer screening with those of a conventional reading system used by radiologists.

DESIGN AND SETTING:

We conducted a comparative study of chest CT images for lung cancer screening at three private institutions.

METHODS:

A database of low-dose (< 2 mSv) chest CT images of patients at risk of lung cancer was viewed with the conventional reading system (301 patients and 226 nodules) or computer-aided diagnosis system without any subsequent radiologist review (944 patients and 1,048 nodules).

RESULTS:

The numbers of detected and solid nodules per patient (both P < 0.0001) were higher using the computer-aided diagnosis system than those using the conventional reading system. The nodule size was reported as the maximum size in any plane in the computer-aided diagnosis system. Higher numbers of patients (102 [11%] versus 20 [7%], P = 0.0345) and nodules (154 [15%] versus 17 [8%], P = 0.0035) were diagnosed with cancer using the computer-aided diagnosis system.

CONCLUSIONS:

The computer-aided diagnosis system facilitates the diagnosis of cancerous nodules, especially solid nodules, in low-dose (< 2 mSv) CT among patients at risk for lung cancer.

KEY WORDS (MeSH terms):
Diagnostic imaging; Early detection of cancer; Lung neoplasms

AUTHORS’ KEY WORDS:
Cancer nodule; Computed tomography; Computer-aided detection system; Image plane; Lung cancer; Radiation dose

Parameters	Predicted categories
No computed tomography images available	0 (Incomplete)
No appearances of nodules in computed tomography images	1 (Absent)
< 6 mm ɸ for solid nodules and part-solid nodules	2 (Benign)
≥ 6 to < 8 mm ɸ for solid nodules and ≥ 6 ɸ with solid component < 6 mm for part-solid nodules	3 (Probably benign)
≥ 8 to < 15 mm ɸ for solid nodules and ≥ 6 ɸ with solid components ≥ 6 mm to < 8 mm for part-solid nodules	4A (Suspicious)
≥ 15 mm ɸ for solid nodules and > 8 ɸ with solid components ≥ 8 mm for part-solid nodules	4B (Very suspicious)
Suspicious nodules with additional features in imaging analysis	4X (Very suspicious)
Significant clinical and imaging parameters	4S (Clinically significant)

Characteristics		Conventional reading system by radiologists	Computer-aided diagnosis system	Comparisons
Numbers of patients included in the analysis		301	944	P value	95% Cl	Df	F value
Sex	Male	281 (93)	873 (92)	0.7032 (Fisher’s test)	0.7424–1.6529	N/A	N/A
Sex	Female	20 (7)	71 (8)	0.7032 (Fisher’s test)	0.7424–1.6529	N/A	N/A
Age (years)		61.15 ± 8.14	62.15 ± 8.55	0.0741 (t-test)	N/A	1243	1.1030
Smoking status	Current	15 (5)	55 (6)	0.8054 (χ²-test)	N/A	2	N/A
	Previous	126 (42)	381 (40)
	None	160 (53)	508 (54)
Participants with available prior computed tomography		51 (17)	141 (15)	0.4099 (Fisher’s test)	0.8636–1.4499	N/A	N/A

Characteristics		Conventional reading system by radiologists	Computer-aided diagnosis system	Comparisons
Total numbers of nodules included in the analysis		226	1,048	P value	95% Cl	F value
Nodule(s)/Patient		0.75	1.11*	< 0.0001 (Fisher’s test)	0.6108 – 0.8256	N/A
Characters of nodules
Solid		180 (80)	943 (90)*	< 0.0001 (Fisher’s test)	0.3994 – 0.6932	N/A
Part-solid		18 (8)	50 (5)	0.0707 (Fisher’s test)	1.0130 – 2.3240	N/A
Pure-ground glass		2 8 (12)*	55 (5)	0.0003 (Fisher’s test)	1.4630 – 2.8150	N/A
Size (mm)
Transverse plane		4.41 ± 1.22	4.32 ± 1.85	0.4846 (t-test)	-0.3425 – 0.1625	2.2990
Maximum orthogonal plane		N/A	4.61 ± 2.05	N/A	N/A	N/A
Any maximum plane		N/A	4.92 ± 1.81*	N/A	N/A	N/A
Comparison for size
P value		N/A	< 0.0001 (ANOVA; F value: 25.9670)	N/A	N/A	N/A
q value	Transverse plane versus maximum orthogonal plane	N/A	4.9250 (95% Cl: -0.4851 – -0.0949)	N/A	N/A	N/A
	Transverse plane versus any maximum plane	N/A	10.1900 (95% Cl: -0.7951 – -0.4049)	N/A	N/A	N/A
	Maximum orthogonal plane versus any maximum plane	N/A	5.2650 (95% Cl: -0.5051 – -0.1149)	N/A	N/A	N/A

Lung-RADS category		Conventional reading system by radiologists	Computer-aided diagnosis system
Lung-RADS category		Transverse plane	Transverse plane	Maximum orthogonal plane	Any maximum plane
Number of patients included in analysis		301	944	944	944
1	Numbers of patients	181 (60)	47 8 (51)	461 (49)	472 (50)
1	Number of patients with cancer	0 (0)	0 (0)	0 (0)	0 (0)
2	Number of patients	91 (30)	352 (37)	363 (38)	368 (39)
2	Number of patients with cancer	7 (2)	18 (2)	25 (3)	42 (4)
3	Number of patients	17 (6)	55 (6)	69 (7)	56 (6)
3	Number of patients with cancer	4 (1)	12 (1)	14 (1)	19 (2)
4A	Number of patients	7 (2)	31 (3)	35 (4)	31 (3)
4A	Number of patients with cancer	5 (2)	21 (2)	23 (2)	28 (3)
4B	Number of patients	3 (1)	11 (1)	12 (1)	13 (1)
4B	Number of patients with cancer	2 (1)	8 (1)	9 (1)	10 (1)
4X	Number of patients	2 (1)	17 (2)	4 (1)	4 (1)
4X	Number of patients with cancer	2 (1)	11 (1)	2 (1)	3 (1)
Total number of patients with cancers		20 (7)	70 (7)	73 (8)	102 (11)
Comparison of total number of patients with cancer		P value			95% Cl
Transverse plane in the conventional reading system versus transverse plane in the computer-aided diagnosis system		0.7032			0.6124–1.3620
Transverse plane in the conventional reading system versus maximum orthogonal plane in the computer-aided diagnosis system		0.6150			0.5901–1.3170
Transverse plane in the conventional reading system versus any maximum plane in the computer-aided diagnosis system		0.0345			0.4333–0.9906
Transverse plane in the computer-aided diagnosis system versus maximum orthogonal plane in the computer-aided diagnosis system		0.8620			0.8214–1.1630
Transverse plane in the computer-aided diagnosis system versus any maximum plane in the computer-aided diagnosis system		0.0130			0.6632–0.9627
Maximum orthogonal plane in the computer-aided diagnosis system versus any maximum plane in the computer-aided diagnosis system		0.0260			0.6844–0.9834

Lung-RADS category		Conventional reading system by radiologists	Computer-aided diagnosis system
Lung-RADS category		Transverse plane	Transverse plane	Maximum orthogonal plane	Any maximum plane
Number of nodules included in analysis		226	1,048	1,048	1,048
2	Number of nodules	189 (84)	905 (86)	864 (82)	785 (75)
2	Number of nodules with cancer	5 (2)	65 (6)	57 (5)	65 (6)
3	Number of nodules	20 (9)	71 (7)	105 (10)	161 (15)
3	Number of nodules with cancer	1 (1)	12 (1)	24 (2)	31 (3)
4A	Number of nodules	8 (4)	45 (4)	32 (3)	52 (5)
4A	Number of nodules with cancer	5 (2)	9 (1)	22 (2)	29 (3)
4B	Number of nodules	7 (3)	18 (2)	36 (3)	35 (3)
4B	Number of nodules with cancer	5 (2)	2 (0.5)	18 (2)	22 (2)
4X	Number of nodules	2 (1)	9 (1)	11 (1)	15 (1)
4X	Number of nodules with cancer	1 (1)	2 (0.5)	5 (1)	7 (1)
Total number of nodules with cancer		17 (8)	90 (9)	126 (12)	154 (15)
Comparison of total number of nodules with cancer		P value		95% Cl
Transverse plane in the conventional reading system versus transverse plane in the computer-aided diagnosis system		0.6921		0.5639–1.396
Transverse plane in the conventional reading system versus maximum orthogonal plane in the computer-aided diagnosis system		0.0622		0.4050–1.022
Transverse plane in the conventional reading system versus any maximum plane in the computer-aided diagnosis system		0.0035		0.3288–0.8373
Transverse plane in the computer-aided diagnosis system versus maximum orthogonal plane in the computer-aided diagnosis system		0.0119		0.6940–0.9633
Transverse plane in the computer-aided diagnosis system versus any maximum plane in the computer-aided diagnosis system		< 0.0001		0.6016–0.8452
Maximum orthogonal plane in the computer-aided diagnosis system versus any maximum plane in the computer-aided diagnosis system		0.0830		0.7727–1.0170

Parameters	CRS by radiologists	CAD			Comparisons
	CRS by radiologists	CAD			P value
	TP (%)	TP	MOP	AMP	TP of CRS vs. TP of CAD	TP of CRS vs. MOP of CAD	TP of CRS vs. AMP of CAD	TP of CAD vs. MOP of CAD	TP of CAD vs. AMP of CAD	MOP of CAD vs. AMP of CAD
Sensitivity	92.69% (68.21–99.81%)	92.61% (67.89–99.11%)	95.87% (70.12–99.15%)	96.15% (71.12–99.85%)	0.8541	0.0612	0.0431	0.0581	0.0411	0.0981
Specificity	89.91% (88.12–92.15%)	88.11% (86.15–91.11%)	87.12% (84.11–89.99%)	82.98% (80.11–88.15%)	0.9121	0.0852	0.0421	0.0651	0.0391	0.0382
PPV	7.52% (5.15–9.15%)	8.59% (7.11–10.12%)	12.02% (10.15–14.11%)	14.69% (11.12–16.52%)	0.6891	0.0611	0.0041	0.0131	< 0.0001	0.0891
NPV	99.81% (98.12–100%)	99.82% (98.08–99.89%)	99.81% (98.15–99.81%)	99.89% (98.86–99.89%)	0.5831	0.6211	0.5541	0.6541	0.6321	0.6641

Associação Paulista de Medicina - APM APM / Publicações Científicas, Av. Brigadeiro Luís Antonio, 278 - 7º and., 01318-901 São Paulo SP - Brazil, Tel.: +55 11 3188-4310 / 3188-4311, Fax: +55 11 3188-4255 - São Paulo - SP - Brazil
E-mail: revistas@apm.org.br

Acompanhe os números deste periódico no seu leitor de RSS

[1] Address for correspondence: Boya Li Department of Medical Imaging, Jiangxi provincial People’s Hospital, The First Affiliated Hospital of Nanchang Medical College, Nanchang 330006, Jiangxi, China. Tel./Fax. +86-0791-86895678 E-mail: l13970818703@sina.com