Automated seeding-based nuclei segmentation in nonlinear optical microscopy

Anna Medyukhina; Tobias Meyer; Sandro Heuke; Nadine Vogler; Benjamin Dietzek; Jürgen Popp

doi:10.1364/AO.52.006979

Applied Optics
Vol. 52,
Issue 28,
pp. 6979-6994
(2013)
•https://doi.org/10.1364/AO.52.006979

Automated seeding-based nuclei segmentation in nonlinear optical microscopy

Anna Medyukhina, Tobias Meyer, Sandro Heuke, Nadine Vogler, Benjamin Dietzek, and Jürgen Popp

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Anna Medyukhina, Tobias Meyer, Sandro Heuke, Nadine Vogler, Benjamin Dietzek, and Jürgen Popp, "Automated seeding-based nuclei segmentation in nonlinear optical microscopy," Appl. Opt. 52, 6979-6994 (2013)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Abstract

Nonlinear optical (NLO) microscopy based, e.g., on coherent anti-Stokes Raman scattering (CARS) or two-photon-excited fluorescence (TPEF) is a fast label-free imaging technique, with a great potential for biomedical applications. However, NLO microscopy as a diagnostic tool is still in its infancy; there is a lack of robust and durable nuclei segmentation methods capable of accurate image processing in cases of variable image contrast, nuclear density, and type of investigated tissue. Nonetheless, such algorithms specifically adapted to NLO microscopy present one prerequisite for the technology to be routinely used, e.g., in pathology or intraoperatively for surgical guidance. In this paper, we compare the applicability of different seeding and boundary detection methods to NLO microscopic images in order to develop an optimal seeding-based approach capable of accurate segmentation of both TPEF and CARS images. Among different methods, the Laplacian of Gaussian filter showed the best accuracy for the seeding of the image, while a modified seeded watershed segmentation was the most accurate in the task of boundary detection. The resulting combination of these methods followed by the verification of the detected nuclei performs high average sensitivity and specificity when applied to various types of NLO microscopy images.

Full Article | PDF Article

More Like This

Automated method for the segmentation and morphometry of nerve fibers in large-scale CARS images of spinal cord tissue

Steve Bégin, Olivier Dupont-Therrien, Erik Bélanger, Amy Daradich, Sophie Laffray, Yves De Koninck, and Daniel C. Côté
Biomed. Opt. Express 5(12) 4145-4161 (2014)

Automated segmentation of retinal pigment epithelium cells in fluorescence adaptive optics images

Piero Rangel-Fonseca, Armando Gómez-Vieyra, Daniel Malacara-Hernández, Mario C. Wilson, David R. Williams, and Ethan A. Rossi
J. Opt. Soc. Am. A 30(12) 2595-2604 (2013)

Multimodal non-linear optical imaging for label-free differentiation of lung cancerous lesions from normal and desmoplastic tissues

Xiaoyun Xu, Jie Cheng, Michael J. Thrall, Zhengfan Liu, Xi Wang, and Stephen T.C. Wong
Biomed. Opt. Express 4(12) 2855-2868 (2013)

Previous Article Next Article

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (12)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (10)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (16)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

	Group	Number of Images	Number of Nuclei (Manually Detected)	Image Resolution, Pix per μm	Image Size, μm ( $W i d t h = H e i g h t$ )	Microscopy Method
1	Brain tumors	18	640	1.609–6.803	47–355	TPEF
2	Larynx	27	626	2.276–4.551	74–228	TPEF
3	Skin tumors	27	533	2.276	158–329	TPEF, CARS
4	Keloidal skin	3	113	1.138	30–47	CARS
	Total	75	1912	1.138–6.803	30–355

	Group	$h = 0$	$h = 20$	$h = 40$	$h = 60$	$h = 80$	$h = 100$
1	Brain tumors	31.6 ( $99.6 / 19.6$ )	62.2 ( $94.4 / 50.0$ )	73.9 ( $85.6 / 69.0$ )	70.8 ( $69.6 / 76.3$ )	63.2 ( $55.7 / 77.1$ )	52.4 ( $42.3 / 75.6$ )
2	Larynx	37.0 ( $99.6 / 24.0$ )	70.6 ( $93.0 / 58.4$ )	78.7 ( $81.5 / 78.0$ )	72.3 ( $65.3 / 84.2$ )	58.1 ( $47.8 / 81.2$ )	45.0 ( $34.6 / 81.0$ )
3	Skin tumors	46.9 ( $99.5 / 31.0$ )	78.3 ( $94.0 / 67.9$ )	83.0 ( $85.1 / 82.2$ )	76.3 ( $71.4 / 84.3$ )	67.1 ( $58.1 / 84.7$ )	54.9 ( $43.7 / 85.5$ )
4	Keloidal skin	43.1 ( $98.3 / 27.7$ )	59.2 ( $95.5 / 43.0$ )	68.0 ( $89.0 / 55.2$ )	73.6 ( $77.2 / 70.5$ )	62.7 ( $58.7 / 67.5$ )	54.0 ( $46.3 / 65.5$ )
	Average	39.5 ( $99.5 / 25.6$ )	70.9 ( $93.8 / 59.2$ )	78.7 ( $84.1 / 76.4$ )	73.4 ( $69.0 / 81.8$ )	62.7 ( $53.8 / 80.9$ )	50.7 ( $40.2 / 80.7$ )

	Group	Sobel, $1 / 2$	Canny, $1 / 2$	Sobel, $1 / 4$	Canny, $1 / 4$
1	Brain tumors	46.5 ( $45.3 / 50.8$ )	45.0 ( $47.5 / 45.6$ )	58.5 ( $61.5 / 59.2$ )	55.5 ( $62.2 / 52.4$ )
2	Larynx	45.2 ( $40.1 / 55.6$ )	44.5 ( $39.5 / 55.0$ )	57.1 ( $52.5 / 64.1$ )	52.0 ( $50.3 / 55.8$ )
3	Skin tumors	78.3 ( $74.9 / 83.0$ )	72.3 ( $72.0 / 73.1$ )	86.1 ( $86.8 / 86.4$ )	81.1 ( $85.9 / 77.8$ )
4	Keloidal skin	68.3 ( $65.7 / 71.4$ )	59.2 ( $65.4 / 54.0$ )	74.6 ( $79.2 / 70.5$ )	67.6 ( $80.0 / 58.7$ )
	Average	58.4 ( $54.9 / 65.0$ )	55.2 ( $54.1 / 59.3$ )	68.6 ( $68.1 / 71.2$ )	63.9 ( $67.2 / 63.0$ )

	Group	$thr = 0.1$ max(R)	$thr = 0.2$ max(R)	$thr = 0.3$ max(R)	$thr = 0.4$ max(R)
1	Brain tumors	79.7 ( $95.0 / 71.4$ )	85.1 ( $94.7 / 78.7$ )	90.6 ( $93.3 / 88.6$ )	89.8 ( $86.1 / 94.4$ )
2	Larynx	87.9 ( $94.4 / 83.2$ )	89.9 ( $94.0 / 86.8$ )	91.1 ( $91.9 / 90.8$ )	89.4 ( $85.1 / 95.0$ )
3	Skin tumors	90.3 ( $96.8 / 85.4$ )	92.4 ( $96.1 / 89.5$ )	92.6 ( $93.2 / 92.4$ )	87.9 ( $83.2 / 94.8$ )
4	Keloidal skin	72.5 ( $97.7 / 57.7$ )	78.0 ( $96.6 / 65.5$ )	83.5 ( $92.7 / 75.9$ )	84.7 ( $83.7 / 85.9$ )
	Average	86.2 ( $95.5 / 80.1$ )	89.2 ( $95.0 / 85.0$ )	91.2 ( $92.8 / 90.2$ )	88.8 ( $84.6 / 94.4$ )

	Group	$h$ -Minima, $h = 40$	CHT, Sobel Edges, $1 / 4$ Doughnut	Scale LoG, $thr = 0.3$ max(R)
1	Brain tumors	73.9 ( $85.6 / 69.0$ )	58.5 ( $61.5 / 59.2$ )	90.6 ( $93.3 / 88.6$ )
2	Larynx	78.7 ( $81.5 / 78.0$ )	57.1 ( $52.5 / 64.1$ )	91.1 ( $91.9 / 90.8$ )
3	Skin tumors	83.0 ( $85.1 / 82.2$ )	86.1 ( $86.8 / 86.4$ )	92.6 ( $93.2 / 92.4$ )
4	Keloidal skin	68.0 ( $89.0 / 55.2$ )	74.6 ( $79.2 / 70.5$ )	83.5 ( $92.7 / 75.9$ )
	Average	78.7 ( $84.1 / 76.4$ )	68.6 ( $68.1 / 71.2$ )	91.2 ( $92.8 / 90.2$ )

	Group	Radial-Gradient-Based Detection	LoG Response	Active Contour	Seeded Watershed
1	Brain tumors	73.3 ( $71.2 / 76.9$ )	76.9 ( $69.9 / 87.0$ )	77.6 ( $71.5 / 86.5$ )	82.9 ( $84.5 / 82.1$ )
2	Larynx	70.2 ( $78.6 / 65.4$ )	75.3 ( $75.9 / 77.3$ )	74.8 ( $75.0 / 77.2$ )	79.2 ( $85.5 / 74.3$ )
3	Skin tumors	70.8 ( $67.0 / 77.2$ )	71.6 ( $62.9 / 87.1$ )	72.8 ( $66.7 / 83.6$ )	80.4 ( $78.9 / 83.0$ )
4	Keloidal skin	64.3 ( $84.8 / 52.2$ )	77.8 ( $82.4 / 73.9$ )	72.8 ( $82.7 / 65.4$ )	75.3 ( $94.4 / 62.9$ )
	Average	71.0 ( $72.9 / 71.9$ )	74.5 ( $70.0 / 83.0$ )	74.7 ( $71.5 / 81.3$ )	80.4 ( $83.3 / 78.8$ )

	Group	Seeding Accuracy (%)	Boundary Accuracy (%)
1	Brain tumors	$93.7 \pm 4.6$ ( $95.3 \pm 4.9 / 92.6 \pm 8.7$ )	$83.0 \pm 3.6$ ( $84.5 \pm 5.3 / 82.2 \pm 7.3$ )
2	Larynx	$94.0 \pm 4.7$ ( $94.7 \pm 5.5 / 94.0 \pm 8.9$ )	$79.2 \pm 4.2$ ( $85.5 \pm 4.9 / 74.3 \pm 7.6$ )
3	Skin tumors	$95.2 \pm 4.4$ ( $95.4 \pm 6.5 / 95.5 \pm 6.7$ )	$80.4 \pm 5.3$ ( $78.5 \pm 9.5 / 83.3 \pm 5.6$ )
4	Keloidal skin	$83.1 \pm 1.39$ ( $93.3 \pm 1.7 / 74.9 \pm 1.7$ )	$75.3 \pm 4.2$ ( $94.4 \pm 1.8 / 62.9 \pm 6.5$ )
	Average	$93.9 \pm 5.0$ ( $95.0 \pm 5.6 / 93.5 \pm 8.8$ )	$80.4 \pm 4.7$ ( $83.1 \pm 7.9 / 79.0 \pm 8.5$ )

	Tissue Type	Dens	${Fr}_{nuc}$	Size	$C$	$T$	$E$	Conv
1	Brain tumors	9.6	17.5	15.9	12.7	3.2	9.5	3.6
2	Larynx	8.9	19.0	23.5	17.3	2.7	11.7	5.2
3	Skin tumors	8.7	12.7	16.7	20.8	7.7	10.1	2.3
4	Keloidal skin	24.5	63.3	39.8	21.1	23.5	13.0	7.8
	Average	9.6	18.2	19.9	17.6	5.5	10.6	3.9

Characteristic	Healthy Brain Tissue	SCC Metastasis Region
Nuclear density per $0.01 {mm}^{2}$	115.5	4.3
Nuclear fraction (%)	27.5	0.9
Size ( ${μm}^{2}$ )	20.6 (14.5–31.3)	20.0 (10.8–28.0)
${Circularity}^{- 1}$	0.21 (0.14–0.27)	0.15 (0.12–0.19)
Texture	0.047	(0.044–0.050)
	0.043	(0.041–0.046)
Eccentricity	2.14 (1.72–2.84)	1.88 (1.66–2.21)
Convexity	0.77 (0.68–0.86)	0.78 (0.74–0.86)

Characteristic	Healthy Epithelium	Dysplastic Epithelium	Connective Tissue
Nuclear density per $0.01 {mm}^{2}$	67.6	84.6	108.9
Nuclear fraction (%)	15.3	27.6	21.4
Size ( ${μm}^{2}$ )	20.7	25.9	15.5
	(14.7–28.5)	(16.3–43.1)	(11.4–23.5)
${Circularity}^{- 1}$	0.24	0.19	0.18
	(0.17–0.30)	(0.14–0.24)	(0.13–0.24)
Texture	0.081	0.082	0.080
	(0.078–0.084)	(0.078–0.085)	(0.076–0.083)
Eccentricity	2.38	2.10	2.04
	(1.97–3.13)	(1.79–2.60)	(1.68–2.59)
Convexity	0.70	0.77	0.83
	(0.61–0.78)	(0.68–0.86)	(0.74–0.88)

Abstract

Cited By

Figures (12)

Tables (10)

Equations (16)

Applied Optics