Limits...
AutoTag and AutoSnap: Standardized, semi-automatic capture of regions of interest from whole slide images.

Marien KM, Andries L, De Schepper S, Kockx MM, De Meyer GR - MethodsX (2015)

Bottom Line: Although SURS is the most reliable sampling method, it implies a high workload.However, SURS can be semi-automated and in this way contribute to the development of a validated quantification method for microvessel counting in the clinical setting.Here, we report a method to use semi-automated SURS for microvessel counting: •Whole slide imaging with Pannoramic SCAN (3DHISTECH)•Computer-assisted sampling in Pannoramic Viewer (3DHISTECH) extended by two self-written AutoHotkey applications (AutoTag and AutoSnap)•The use of digital grids in Photoshop(®) and Bridge(®) (Adobe Systems) This rapid procedure allows traceability essential for high throughput protein analysis of immunohistochemically stained tissue.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Physiopharmacology, University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgium ; HistoGeneX NV, ZNA Middelheim Campus, Lindendreef 1, B-2020 Antwerp, Belgium.

ABSTRACT
Tumor angiogenesis is measured by counting microvessels in tissue sections at high power magnification as a potential prognostic or predictive biomarker. Until now, regions of interest (ROIs) were selected by manual operations within a tumor by using a systematic uniform random sampling (SURS) approach. Although SURS is the most reliable sampling method, it implies a high workload. However, SURS can be semi-automated and in this way contribute to the development of a validated quantification method for microvessel counting in the clinical setting. Here, we report a method to use semi-automated SURS for microvessel counting: •Whole slide imaging with Pannoramic SCAN (3DHISTECH)•Computer-assisted sampling in Pannoramic Viewer (3DHISTECH) extended by two self-written AutoHotkey applications (AutoTag and AutoSnap)•The use of digital grids in Photoshop(®) and Bridge(®) (Adobe Systems) This rapid procedure allows traceability essential for high throughput protein analysis of immunohistochemically stained tissue.

No MeSH data available.


Related in: MedlinePlus

Calculation of the minimum number of regions of interest (ROIs) required for analysis of microvessel density. Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (see Fig. 2). Random sampling with replacement (i.e. bootstrapping) was carried out 1000 times for the calculation of microvessel densities (QA, expressed as number of microvessels per area). The coefficient of variation (CV) was plotted versus the number of ROIs (f(x) = 0.4446x−0.42, R2 = 0.99). The first derivative of this function (f′(x) = −0.1867x−1.42) was used to define the minimum number of ROIs required. This value was reached when the difference between two consecutive local derivatives was smaller than 0.50% (in the example shown, the minimum number of ROIs required was six).
© Copyright Policy - CC BY
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4487922&req=5

fig0015: Calculation of the minimum number of regions of interest (ROIs) required for analysis of microvessel density. Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (see Fig. 2). Random sampling with replacement (i.e. bootstrapping) was carried out 1000 times for the calculation of microvessel densities (QA, expressed as number of microvessels per area). The coefficient of variation (CV) was plotted versus the number of ROIs (f(x) = 0.4446x−0.42, R2 = 0.99). The first derivative of this function (f′(x) = −0.1867x−1.42) was used to define the minimum number of ROIs required. This value was reached when the difference between two consecutive local derivatives was smaller than 0.50% (in the example shown, the minimum number of ROIs required was six).

Mentions: The minimum number of regions of interest (ROIs) required for analysis of microvessel density was calculated by using random sampling with replacement (bootstrapping; Fig. 3). Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (Fig. 2). Microvessel densities (QA = ΣiNi/ΣiVi,ref with an i value from 1 to 25 ROIs, expressed as number of microvessels per area, with Ni the number of counted vessels in ROI i and Vi,ref the number of grid points hitting tissue in ROI i) were calculated 1000 times by resampling from the pool of 25 ROIs. The coefficient of variation (CV) was plotted versus the number of ROIs (Fig. 3). A trend line of the form f(x) = ax−b was added and the first derivative of this function (f′(x) = −abx−b−1) was used to define the minimum number of ROIs required. The local derivative must be smaller than 0.50%. After creating these graphs (Fig. 3) for 19 colorectal carcinomas, 22 renal cell carcinomas, 21 glioblastomas, and 21 ovarian carcinomas, we concluded that 10 regions of interest are sufficient for accurate microvessel density measurements.


AutoTag and AutoSnap: Standardized, semi-automatic capture of regions of interest from whole slide images.

Marien KM, Andries L, De Schepper S, Kockx MM, De Meyer GR - MethodsX (2015)

Calculation of the minimum number of regions of interest (ROIs) required for analysis of microvessel density. Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (see Fig. 2). Random sampling with replacement (i.e. bootstrapping) was carried out 1000 times for the calculation of microvessel densities (QA, expressed as number of microvessels per area). The coefficient of variation (CV) was plotted versus the number of ROIs (f(x) = 0.4446x−0.42, R2 = 0.99). The first derivative of this function (f′(x) = −0.1867x−1.42) was used to define the minimum number of ROIs required. This value was reached when the difference between two consecutive local derivatives was smaller than 0.50% (in the example shown, the minimum number of ROIs required was six).
© Copyright Policy - CC BY
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4487922&req=5

fig0015: Calculation of the minimum number of regions of interest (ROIs) required for analysis of microvessel density. Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (see Fig. 2). Random sampling with replacement (i.e. bootstrapping) was carried out 1000 times for the calculation of microvessel densities (QA, expressed as number of microvessels per area). The coefficient of variation (CV) was plotted versus the number of ROIs (f(x) = 0.4446x−0.42, R2 = 0.99). The first derivative of this function (f′(x) = −0.1867x−1.42) was used to define the minimum number of ROIs required. This value was reached when the difference between two consecutive local derivatives was smaller than 0.50% (in the example shown, the minimum number of ROIs required was six).
Mentions: The minimum number of regions of interest (ROIs) required for analysis of microvessel density was calculated by using random sampling with replacement (bootstrapping; Fig. 3). Twenty-five ROIs were chosen in a CD31-stained colorectal carcinoma slide (Fig. 2). Microvessel densities (QA = ΣiNi/ΣiVi,ref with an i value from 1 to 25 ROIs, expressed as number of microvessels per area, with Ni the number of counted vessels in ROI i and Vi,ref the number of grid points hitting tissue in ROI i) were calculated 1000 times by resampling from the pool of 25 ROIs. The coefficient of variation (CV) was plotted versus the number of ROIs (Fig. 3). A trend line of the form f(x) = ax−b was added and the first derivative of this function (f′(x) = −abx−b−1) was used to define the minimum number of ROIs required. The local derivative must be smaller than 0.50%. After creating these graphs (Fig. 3) for 19 colorectal carcinomas, 22 renal cell carcinomas, 21 glioblastomas, and 21 ovarian carcinomas, we concluded that 10 regions of interest are sufficient for accurate microvessel density measurements.

Bottom Line: Although SURS is the most reliable sampling method, it implies a high workload.However, SURS can be semi-automated and in this way contribute to the development of a validated quantification method for microvessel counting in the clinical setting.Here, we report a method to use semi-automated SURS for microvessel counting: •Whole slide imaging with Pannoramic SCAN (3DHISTECH)•Computer-assisted sampling in Pannoramic Viewer (3DHISTECH) extended by two self-written AutoHotkey applications (AutoTag and AutoSnap)•The use of digital grids in Photoshop(®) and Bridge(®) (Adobe Systems) This rapid procedure allows traceability essential for high throughput protein analysis of immunohistochemically stained tissue.

View Article: PubMed Central - PubMed

Affiliation: Laboratory of Physiopharmacology, University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgium ; HistoGeneX NV, ZNA Middelheim Campus, Lindendreef 1, B-2020 Antwerp, Belgium.

ABSTRACT
Tumor angiogenesis is measured by counting microvessels in tissue sections at high power magnification as a potential prognostic or predictive biomarker. Until now, regions of interest (ROIs) were selected by manual operations within a tumor by using a systematic uniform random sampling (SURS) approach. Although SURS is the most reliable sampling method, it implies a high workload. However, SURS can be semi-automated and in this way contribute to the development of a validated quantification method for microvessel counting in the clinical setting. Here, we report a method to use semi-automated SURS for microvessel counting: •Whole slide imaging with Pannoramic SCAN (3DHISTECH)•Computer-assisted sampling in Pannoramic Viewer (3DHISTECH) extended by two self-written AutoHotkey applications (AutoTag and AutoSnap)•The use of digital grids in Photoshop(®) and Bridge(®) (Adobe Systems) This rapid procedure allows traceability essential for high throughput protein analysis of immunohistochemically stained tissue.

No MeSH data available.


Related in: MedlinePlus