Update README.md

README.md

@@ -50,12 +50,12 @@ IsoNet is kind of conservative in adding information into missing wedge region.
 However, there are some ways to increase your success rate.
 1. IsoNet performs better in high contrast tomograms. That means it will be helpful to tweak the parameters (especially snrfalloff) in CTF deconvolution step to make increase the weight of low resolution information. Or trying with the data acquired with phaseplate first. As far as we know, phaseplate data will always give you good result.
 
-2. Missing wedge caused the nonlocal distributted information. You may observed the long shadows of gold beads in the tomograms, and those long shadows can not be fully corrected with sub-tomogram based missing correction in IsoNet, because the receptive field of the network is limitted to your subtomogram. This nonlocal information makes it particular difficult to recover the horizontal oriented membrane. There are several ways to improve. **First**, training with tomograms with larger size, the default cube size is 64, you may want to increase the size to 96 or 128, however this may lead to the OOM error Please refer to FAQ #1 when you have this problem. **Second**, bin your tomograms more. Some times we even bin our celluar tomograms to 20A/pix for IsoNet processing, this will of course increase your network receptive field, given the same size of subtomogram. 
+2. Missing wedge caused the nonlocal distributted information. You may observed the long shadows of gold beads in the tomograms, and those long rays can not be fully corrected with sub-tomogram based missing correction in IsoNet, because the receptive field of the network is limitted to your subtomogram. This nonlocal information makes it particular difficult to recover the horizontal oriented membrane. There are several ways to improve. **First**, training with tomograms with larger size, the default cube size is 64, you may want to increase the size to 96 or 128, however this may lead to the OOM error Please refer to FAQ #1 when you have this problem. **Second**, bin your tomograms more. Some times we even bin our celluar tomograms to 20A/pix for IsoNet processing, this will of course increase your network receptive field, given the same size of subtomogram. 
 
 3. IsoNet is currently designed to correct missing wedge for tomograms with -60 to 60 degress tilt range. The other tilt scheme or when the tomograms have large x axis tilt. The results might not be optimal. 
 ## 4. Can not create a good mask during mask generation step
 The mask is only important if the sample is sparsely located in the tomograms. And the mask do not need to be perfect to obtain good result, in other words, including many empty/unwanted subtomos during the refinement can be toralated. 
 
 To obtain a good mask, the tomograms should have sufficient contrast, which can be achieved by CTF deconvolution. User defined mask can also be supplied by changing the mask_name field in the star file. Alternately, you can also use subtomograms extracted with other methods and skip the entire mask creation and subtomograms extraction steps.
 
-If you want to exclude carbon area of the tomograms, you can try the new mask boundary feature in version 0.2. It allows you to draw a polygon in 3dmod so that the area outside the polygon will be excluded.
+If you want to exclude carbon area of the tomograms, you can try the new mask boundary feature in version 0.2. It allows you to draw a polygon in 3dmod so that the area outside the polygon will be excluded.